RADIOLOGY 0NCOLO.; o]TI. 1996 Vol. 30 No. 1 Ljubljana ISSN 1318-2099 UDC 616-006 CODEN: RONCEM AGFA VAŠ RAZVOJNI PUT U BUDUCNOST PUNU NAJRAZLICITIJIH MOGUCNOSTI Agfa dijagnosticki centar za kompjuteri­ziranu radiologiju Prekretnica u CR kva­liteti slike -maksi­malna produktivnost putem jedinstvenih kasetnih bufera - pravi kljuc za konven­ cionalnu radiologiju, za potpuno elektron­ sku okolinu buducno­ sti. Agfa lmpax Drystar Najsuvremeniji sustav suhih snimaka -skala u boji i/ili siva -kvali­teta slike zasnovana na dokazanoj tehnolo­giji -spremnost za trenutnu integraciju u lmpax mrežu. Agfa nudi pun i raznolik asortiman za svaku primjenu snimanja. Naša roba može se koristiti s višestrukim sustavima, naši su digitalni su­stavi DICOM kompatibilni, a svi naši sustavi orijentirani su prema budu6nosti. Partnerstvo s tvrtkom AGFA sada, donosi Vam maksi­malnu produktivnost, a istovremeno Vas po­stavlja na put, koji ima najve6e razvojne mo­gu6nosti i vodi k otvorenoj okolini u budu6nosti. Agfa kompaktni centar za snimanje Koristi konvencionalnu kasetu za dnevno svjetlo, filmove bez kaseta i snimke s digitalnih izvora -pre­kretnica u kombinaciji konvencio­nalnog i digitalnog snimanja. Agfa lmpax sustavi za umreža­vanje Otvoreni razvojni put ka kom­pletnoj komunikaciji i upravlja­nju podacima iz bilo kojeg iz­vora za snimanje -DICOM kompatibilnog, ukljucujuci kon­ vencionalnu radiologiju. AGFA-GEVAERT Ges.m.b.H. A-1153 Wien Mariahilfer Str. 198 Tel.: 43-1/891-12 Fax: 43-1/891-12 BAYER AG Predstavništvo Zagreb AGFA-ODJEL 10000 ZAGREB Ivana Luci6a 6 Tel.: 385-1/6126-924 Fax: 385-1/539-304 Osigurajte svoje buduce investicije -AGFA je pravi izbor! RADIOLOGY AND ONCOLOGY Radiology and Oncology is a journal devoted to publication of original contributions in diagnostic and interventional radiology, computerized tomography, ultrasound, magnetic resonance, nuclear medicine, radiotherapy, clinical and experimental oncology, radiophysics and radiation protection. Editor in chief Tomaž Benulic Ljubljana, Slovenia Associate editors Gregor Serša Ljubljana, Slovenia Viljem Kovac Ljubljana, Slovenia Editorial board Tullio Giraldi Branko Palcic Udine, Italy Vancouver, Canada Marija Auersperg Ljubljana, Slovenia Andrija Hebrang Zagreb, Croatia Jurica Papa Zagreb, Croatia Haris Boko Durtla Horvat Zagreb, Croatia Zagreb, Croatia Dušan Pavcnik Nataša V. Budihna Ltiszlo Horvtith Ljubljana, Slovenia Ljubljana, Slovenia Pecs, Hungary Stojan Plesnicar Malte Clausen Berta Jereb Ljubljana, Slovenia Kiel, Germany Ljubljana, Slovenia Ervin B. Podgoršak Christoph Clemm Vladimir Jevtic Montreal, Canada Milnchen, Germany Ljubljana, Slovenia Mario Corsi Udine, /taly H. Dieter Kogelnik Salzburg, Austria Jan C. Roos Amsterdam, The Netherlands Christian Dittrich Vienna, Austria Ivan Lovasic Rijeka, Croatia Horst Sack Essen, Germany Slavko Šimunic Ivan Drillkovic Zagreb, Croatia Marijan Lovrencic Zagreb, Croatia Zagreb, Croatia Gillian Duchesne Luka Milas Lojze Šmid London, Great Britain Houston, USA Ljubljana, Slovenia Bela Pomet Maja Osmak Andrea V eronesi Budapest, Hungary Zagreb, Croatia Gorizia, Italy Publishers Slovenian Medica[ Society -Section of Radiology, Section of Radiotherapy Croatian Medica/ Association -Croatian Society oj Radiology Affiliated with Societas Radiologorum Hungarorum Friuli-Venezia Giulia regional groups of S.I.R.M. (Italian Society of Medica! Radiology) Correspondence address Radiology and Oncology Institute of Oncology Vrazov trg 4 I000 Ljubljana Slovenia Phone: + 386 6I l 320 068 Fax: +386 61 1314 180 Reader Jor English Olga Shrestha Design Monika Fink-Serša Key words and UDC Eva Klemencic Secretaries Milica Harisch Betka Savski Printed by Tiskarna Tone Tomšic, Ljubljana, Slovenia Published quarterly Bank account nurnber 50101 678 48454 Foreign currency account number 50100-620-133-27620-5 l 30/6 LB Ljubljanska banka d.d. -Ljubljana Subscription fee for institutions 100 USD, individuals 50 USD. Single issue for institutions 30 USD, individuals 20 USD. According to the opinion of the Government of the Republic of Slovenia, Public Relation and Media O.ffice, the journal RADIOLOGY AND ONCOLOGY is a publication of inforrnative value, and as such subject to taxation by 5 % sales tax. lndexed and abstracted by: BIOMEDICINA SLOVENlCA CHEMICAL ABSTRACTS DV/7DDD'T'A A,fDT)T/'AIDT D/'-,TDn'l\TTr DTTDT TC'lJT"h.Tf"" r\TTTTC'Tlll\.l CONTENTS ULTRASOUND Ultrasonically guided fine-needle aspiration biopsy in early detection of regional lymph-node involvement in patients with primary planocellular (squamous-cell) carcinoma of oral cavity Šustic A, Žgaljardic Z, Car M, Fuckar Ž, Brumini D, Juretic M S Duplex-Doppler ultrasound of intrarenal arteries in the assessment of the percutaneous transluminal angioplasty of renal artery stenosis -a case report Brkljacic B, Hebrang A, Drinkovic 1 NUCLEAR MEDICINE Uptake kinetics of radiolabelled 1,8 dihydroxyanthraquinone and acridinone derivatives in cultures of breast cancer cells Bohuslavizki KH, Wolf H, Kutzner D, Brenner W, Tinnemeyer S, Sippel C, Clausen M, Henze E EXPERIMENTAL ONCOLOGY Incidence, Iatency period, and survival of mice bearing 20-methyl cholantrene induced tumours do not change after Cyclosporin treatment Kotnik V, Banic S CLINICAL ONCOLOGY Risk for recurrence of intracranial germinoma Strojan P, Popovic M, Petric-Grabnar G, Župancic N, Jereb B 25 Carinal resection and reconstruction double-barrelled type for bronchogenic and metastatic carcinoma Vladovic-Relja T, Slobodnjak Z, Majeric-Kog/er V, Karadža J, Gorecan M 32 Pneumonitis after bronchoplastic surgery in stage I lung cancer Kouzmine IV, Khartchenko VP Pneumatosis intestini and pneumoperitoneum in acute lymphoblastic leukemia Frkovic M, Mandic A, Jonic N, Labar B, Mrsic M 41 Causes of fertile disturbances in oncological male patients Kovac V 46 EPIDEMIOLOGY The role of phases of the moon in the development of spontaneous pneumothorax Sok M, Eržen J RADIOPHYSICS A simplified technique for aligning radiation fields in portal imaging Wang H, Fallone BG REPORTS The 2nd Asian-Pacific congress of cardiovascular and interventional radiology Horvdth L 66 Tibor Huth Urology days in Pecs, Hungary Palk6 A 68 NOTICES 69 Radio! Oncol 1996; 30: 5-8. Ultrasonically guided fine-needle aspiration biopsy in early detection of regional lymph-node involvement in patients with primary planocellular (squamous-cell) carcinoma of oral cavity Alan Šustic, 1 Zoran Žgaljardic, 2 Marijan Car, 2 Željko Fuckar, 1 Diego Brumini, 1 Mirna Juretic2 1 Ultrasound Unit, Clinical Hospital Rijeka 2 Department of Maxillo-facial Surgery, Clinical Hospital Rijeka, 51000 Rijeka, Croatia The authors present personal experience in fine-needle aspiration biopsy under ultrasonic guidance (USGFNAB) of neck lymph-nodes in 63 patients with primary planocellular (squamous) carcinoma of oral cavity. Punctured, single, ipsilateral lymph-nodes were not larger than 3 cm of diameter (NI), while the indication far USGFNAB was established on the basis of at least one of the following ultrasonographic parameters: l. disruption of capsular continuity of lymph-nodes, 2. the longest diameter above 20 mm, 3. the relation of short axis to long axis (of lymph-node) higher than 0.6, 4. the absence of intranodal linear echogenic structures. Sensitivity rate of USGFNAB was as high as 88.2 %, specificity 91. 7%, accuracy 88.9 %, respectively. The authors stress the value of ultrasonographic image in conjunction with such harmless and simple method in the detection of occult neck metastases. Key words: mouth neoplasms; carcinoma, squamous cele; lymph node-ultrasonography; biopsy, needle Introduction The assessment of neck lymph-nodes is an integral part of diagnostic procedure in patients with carcinoma of oral cavity, because eventual metastatic deposits thoroughly change thera­peutic management and present significant pre- Corrcspondencc to: mr. se. dr. Alan Šustic, Ultra­sound Unit, Clinical Hospital Rijcka, 51000 Rijeka, Croatia. Phone: +385 51 21 68 99/217. Fax: +385 51 25 80 38. UDC: 616.31-006.6:616-428-033.2-076 dictive element. Modem diagnostics of neck metastases include palpation, conventional colo-Doppler1 and interventional sonography, computed tomography (CT) and magnetic reso­ 3 nance (MR).2' Recent comparative studies clearly show that ultrasonically guided, fine­needle aspiration biopsy (USGFNAB) is the method of choice in detecting metastatic masses in neck lymph-nodes.4• 5 The purpose of the presented study is the evaluation of this method in diagnosis of meta­stasis in a single, ipsilateral lymph node, 3 cm or less in greatest dimension (Nl) in patients Šustic A et al. with planocellular (squamous-cell) carcinoma of oral cavity. Subjects and methods In the period from l. april to l. october 1993 (18 months) USGFNAB was performed on ultrasonographically suspect neck lymph-nodes in 63 patients with pathohistologically verified planocellular (squamous-cell) carcinoma of oral cavity. The age-range was between 29 to 75 years ( average 59 years) with male to female ratio 47:16 (75% : 25%) in favor of male patients. The indication for USGFNAB was based on at least one of the following sonogra­ 7 phic parameters:6 • l. hypoechogenic disruption of capsular con­tinuity of lymph-nodes, 2. the longest diameter above 20 mm, 3. the relation of short axis to long axis ( of lymph-node) higher than 0.6 (SA/LA > 0.6) and 4. the absence of intranodal linear echogenic structures ( stria). With fine-needle (22 gauge) lymph-nodes from 0.6 to 3.0 cm of size were punctured and each lymph-node was punctured only once. The patients with more than one positive node, as well as those with nodes larger than 3 cm were excluded from the study. The instrument used was Aloka SSD-280 LS and linear sonde 5 MHz. Cytologic material, stained according to stan­dard technique (May-Grunvald-Giemsa), was analyzed under small and middle enlargement, and under immersed objective. The cases with positive cytologic findings underwent surgical and oncologic treatment, while those with negative results are regularly controlled clinically and sonographically. Results On Table 1 results of USGFNAB and pathohy­stologic findings in lymph-nodes are presented. USGFNAB revealed metastatic masses in neck lymph-nodes in 46 cases, with one false-positive result. False-negative findings were present in 6 from 17 patients who underwent regular cyto­logic examination. The percent of USGFNAB sensitivity on the presented material was 88.2 % , specificity 91. 7 % , accuracy 88.9 % , positive predictive value 97 .8 % and negative predictive value 64. 7 % , respectively. In 16 cases (25.4 % ) with positive cytologic and pathohistologic findings, occult, impalpable metastases of neck were diagnosed. The preci­sed indication for USGFNAB based on conven­tional sonographic image was exact in 51 (81 % ) cases. There were no complications in any case, except, slight, transitory pain. Table l. Ultrasonically guided, finc-nccdle aspiration biopsy (USGFNAB) of neck lymph-nodes. Histology or Clinical/Ultrasonography Course USGFNAB Malignant Normal Discussion The determination of metastatic deposits in neck lymph-nodes in patients with planocellular (squamous-cell) carcinoma of oral cavity and their early detection is crucial for selection of correct subsequent therapy. Conventional high­ly-resolute ultrasonography of the neck has high sensitivity, but is not specific enough, 2 so that recent studies recommend additional cyto­ 9 logic aspiration. 8• The results of USGFNAB of neck lymph-no­des are presented. Punctured, single ipsilateral nodes were not larger than 3 cm in diameter (Figures 1, 2), while the indication for aspira­tion biopsy was based on conventional sono­graphic examination and following at least one of aforementioned parameters: l. hypoechogenic disruption of capsular con­tinuity of lymph nodes, 2. the longest diameter above 20 mm, 3. the relation between short to long axis ( of Ultrasonically guided fine-needle aspiration biopsy After pathohistologic verification, i.e. regular clinical and ultrasonographic follow-up, we were able to discover metastatic deposits in 51 (81 % ) patients, which signifies that correct in­dication for USGFNAB occurred after conven­tional ultrasonographic examination in 4/5 of patients. In 16 cases (25.4 % ) occult metastatic deposits were found, i.e. lin 31.4 % from 51 patients, regional Nl metastases were not pal­pable. USGFNAB had positive predictive value 97 .8 % and negative predictive value 64. 7 % . By correlating sensitivity and specificity of this method with recent studies of Siegert et al.9 (sensitivity 90 % ) and Baatenburg De Jung and Westerhot-8 ( specificity 92. 9 % ) we can notice somewhat lower sensitivity and specificity rate. Such slight difference (88.2 % vs 90 % ; 91.7 % vs 92.9 % : p = NS) could be explained with the fact that we didn't treat cases with lymph-node metastases larger tl1an 3 cm, that our testing­group was much smaller, and also with the grade of experience of specialist in cytology. Conclusion Ultrasonically guided, fine-needle aspiration biopsy is the method of choice in neck-staging of patients with planocellular (squamous-cell) carcinoma of oral cavity. Sensitivity, specificity and accuracy rate of this method is significantly greater tl1an palpation, conventional ultrasono­graphy, CT or MR2 · 3 · 8 and that independently of the size of metastatic masses in neck lymph­nodes.10 Moreover, this technique puts little stress on the patient, anesthesia is not required and lastly the cost is relevantly lower than CT or MR.9 Summing personal experience and data from recent literature, we suggest that all patients with planocellular (squamous-cell) of oral cavity should undergo conventional ultraso­nographic examination of the neck, and that each suspect sonographic finding should be complemented with USGFNAB. Presented po­sitive and negative predictive value advocate further clinical and ultrasonographical follow­up, also in patients with regular findings. Šustic A et al. References l. Juretic M, Šustic A, Fuckar ž, Car M. Evaluation of metastatic invasion in the wall of main neck vessels. Conventional vs color-coded ultrasono­graphy. Radio! Oncol 1992; 26: 304-7. 2. Leicher-Duber A, Bleicr R, Duber C, Thelen M. Regional lymph-node mctastases in malignat tu­mors of the head and neck: value of diagnostic procedures. Laryngorhinootologie 1991; 70: 27­31. 3. Quetz JU, Rohr S, Hoffmann P, Wustrow J, Mertens J. B-image sonography in lymph-node staging of the head and neck area. A comparasion with palpation, computerized and magnetic reso­nance tomography. HNO 1991; 39: 61-3. 4. Gupta S, Gupta RK, Gujral RB. Ultrasound guided fine-nedle aspiration of nasophraryngeal mass. J Ciin V/tras 1993; 21: 350-1. 5. Baatenburg de Jung RJ, Rongen RJ, Verwoerd CD, Van Overhagen H, Lameris JS, Knegt P. Ultrasound-guided fine-needle aspiration biopsy of neck nodes. Arch Otolaryngol Head Neck Surg 1991; 117: 402-4. 6. Rainer T, Ofner G, Marckhgott E. Ultrasound diagnosis of regional Iymph-node metastass of the neck in patients with head-neck neoplasms: sono­morphologic criteria and diagnostic accuracy. La­ryngorhinootologie 1993; 72: 73-7. 7. Stiglich F, Barbonetti C, Di Lorenzo E, Gherardi G, Maspero S, Bottinelli O, Bonomo F, Bottinelli G, Campani R. Diagnostic reliability of ultrasono­graphy in head and neck neoplasm. Radio/ Med 1991; 81: 838-43. 8. Baatenblurg de Jong RJ, Westerhof JP. Ultra­sound imaging with ultrasound-guided, fine-needle aspiration biopsy in assessment of limph-node metastasis in the neck. Medica/ Care 1993; 5-6: 27-9. 9. Siegert R, Kupers P, Barreton G. Ultrasonogra­phic fine-needle aspiration of pathological masses in the head and neck region. J Ciin Ultrasound 1992; 20: 315-20. 10. Van der Brekel MW, Castelijus JA, Stel HV, Luth WJ, Valk J, Van der Vaal I, Snow GB. Occult metastatic neck disease: detection with US and US-guided fine-needle aspiration cytology. Radiology 1991; 180: 457-{il. Radio/ Oncol 1996; 30: 9-13. Duplex-Doppler ultrasound of intrarenal arteries in the assessment of the percutaneous transluminal angioplasty of renal artery stenosis -A case report Boris Brkljacic, 1 Andrija Hebrang, 2 Ivan Drinkovic1 1 Ultrasonic Center and 2 Interventional Division, Department of Radiology, University Hospital Merkur, Medica! School of the University of Zagreb, Croatia A case is presented where the color duplex-Doppler sonography of intrarenal arteries was used for the assessment of success of percutaneous transluminal angioplasty of renal artery stenosis. Spectral waveform analysis from intrarenal arteries in a case of angiographically praven renal artery stenosis, with the diameter reduction > 80 %, has shown spectra with prolonged systolic acceleration tirne and increased diastolic flow (parvus and tardus type). After the angioplasty had been performed intrarenal arterial spectra have returned to normal forms, with normal systolic acceleration tirne and normal diastolic flow. Duplex-Doppler spectral analysis can be used for the assessment of the outcome of percutaneous transluminal angioplasty in cases of hemodinamically significant renal artery stenosis. Key words: renal artery obstruction; angioplasty, balloon; renal artery -ultrasonography; ultraso­nography, Doppler, duplex Introduction Hemodynamically significant renal artery steno­sis (RAS) causes activation of the renin-angio­tensin system and leads to the renovascular hypertension with deleterious effects on cardio­vascular system and the kidney .1 Percutaneous transluminal angioplasty (PTA) is a preferred noninvasive means of the treatment of RAS, which obviates the need for surgery, thus re­ducing surgical and anesthetical risks, as well Correspondence to: Dr. se. Boris Brkljacic, Ultrasonic Center, Department of Radiology, University Hospital »Merkur«, Zajceva 19, 10000 Zagreb, Croatia; Tel/Fax + 385-1-2331-440. UDC 616.136.7-007.272-089;534-8 as the price of the treatment.2 Duplex-Doppler sonography of main renal arteries has several significant drawbacks in the detection of 4 RAS.3• However, duplex-Doppler of intrare­nal arteries is much easier and quicker to per­form5 and this method may have a significant role in detection, as well as evaluation of the­rapy of RAS. We present a patient who was examined by color-duplex Doppler ultrasound · (CDD US) of intrarenal arteries to evaluate the success of the PT A of renal artery in a case of a hemodinamically significant RAS. Patients and methods A 42 years old male patient with a history of hypertension, lasting for several months, with Brkljacic B et al. maximal blood pressure values of 220/ 130 mmHg. Complete clinical work-up was per­formed, and, eventually, the patient was sub­mitted to conventional selective renal angio­graphy with renin measurements. These results showed significant left renal artery stenosis, with the reduction of the vessel diameter > 80 % . Patient was referred to our hospital for PTA of the left renal artery. One day before the procedure patient was examined by color duplex-Doppler ultrasound (CDD US) of renal and intrarenal arteries. One day after the PTA was performed the patient was reexamined by CDD US. Conventional US and CDD US of both kid­neys was performed with a color-Doppler scan­ner (Radius CF, GE-CGR, Buc, France) with a curved array 3.75 MHz transducer. The pa­tient was examined in both instances by the first author. Examinations were performed in supine and semioblique positions, and the dura­tion of each examination was around 20 minu­tes. Intrarenal interlobar and arcuate arteries were insonated in the upper and the lower pole of the kidney, sample vol ume positioned with help of the color, and spectral waveform analy­sis performed. In insonated arteries, form of spectra was evaluated qualitatively, regarding systolic acceleration tirne and diastolic flow. Main renal arteries could not have been ade­quately visualized due to the excessive arnount of air in bowels. Only the small segment of right main renal artery was analyzed, and left renal artery was not visualized at ali. The wall filter was set usually at 50 Hz, and the sample vol ume was 1-3 mm wide. Hard-copies of spect­ral waveforms were obtained by a Hitachi color­video printer. Prior to PTA, selective digital subtraction angiography (DSA) of the left renal artery was performed on a Siemens DSA machine. After the selective catheterization of the left renal artery, a teflon coated 0.035" guidewire of 140 cm length (Angiomed) was introduced. A 5 .5 F balloon catheter (Angiomed) with a bal­loon diarneter of 6 millimeters was introduced over the guidewire. The balloon was filled three times under the pressure of 7 atm. During the angioplasty the guidewire was left in the segmenta! renal artery branch. lmmediately af­ter the angioplasty a control angiography was perforrned. The duration of the procedure was 20 minutes. Post-procedure compression of the puncture site of the common femoral artery was performed for 10 minutes. Results CDDUS of intrarenal arteries performed one day before angioplasty showed spectra with prolonged systolic acceleration tirne and increa­sed diastolic flow in the left kidney. A name "parvus and tardus waveform" has been propo­sed by Stavros for this type of spectral wave­forms. 6 The diastolic flow was increased compa­red to normal intrarenal waveforms. In the right kidney, which was angiographically nor­mal, the normal morphology of spectra was noted in intrarenal arteries, with sharp systolic rise and short systolic acceleration tirne, and with continuous antegrade diastolic flow. The spectrum from the intrarenal arteries of the left kidney with RAS is shown in the Fi­gure l. Du.plex-Doppler ultrasound of inlrarenal arteries The selective DSA of the left renal artery confirmed conventional angiographic finding of a high-grade (> 80 % ) stenosis in the proximal segment of the left rnain renal artery. The selective DSA of the left renal artery prior to the perforrnance of the PT A is shown in the Figure 2. _,. __ ,. __ A,R! tub 'l Figure 2. The selective DSA of the left renal artery showing a high-grade (> 80 % ) stenosis in the proximal segment of the left main renal artery. The PTA of the left renal artery was techni­cally successful, and the angioplasty was perfor­med without any cornplications during or after the procedure. The irnmediate post-PTA selec­tive DSA of left renal artery is shown in the Figure 3. It showed normal arterial lumen with only minor marginal filling defect. Thirty hours after the angioplasty had been performed CDDUS of intrarenal arteries sho­wed normal forrns of intrarenal spectra, with sharp systolic rise and normal systolic accelera­tion tirne, with reduction of diastolic flow as cornpared to pre-procedure spectra. Post-proce­dure normal spectra in intrarenal arteries are shown in the Figure 4. PTAA,I Figure 3. The selective DSA of thc left renal artery immediately after the PTA was performed. The arte­rial lumen was almost completely restored, with only minor marginal filling defect. Figure 4. US image of the normal intrarenal spectra, with normal pulsatility, sharp systolic rise and normal acceleration tirne 30 hours after the angioplasty had been performed. Discussion The color duplex-Doppler US has enabled the visualization of flow in intrarenal vessels, and has considerably facilitated the performance of spectral waveform analysis and the quantifica­ Brkljacic B et al. 5 tion of Doppler signals from these vessels.4• Duplex-Doppler US of intrarenal arteries has been extensively used in the last decade for evaluation of flow in transplanted kidneys and native kidneys, for diagnosis of renal obstruc­tion, renal vein thrombosis, differentiation of benign and malignant renal masses and in diag­nosis of various parenchymal kidney disea­ s, 7-11 ses. Severa) papers have been published about the usage of duplex-Doppler US in the diagno­sis of renal artery stenosis, and the 0.05). - Latency periods after tumour lnductlon (days) Figure l. Latency periods of 20-methyl cholantrene induced tumours in the controls, 250 f.Lg/dose and 1000 f.Lg/dose CsA treated groups of mice. The first tumours appeared on Day 56 and the last ones on Day 161. The average latency period for the control group (-•-) was 85.94, for the group treated with 250 f.Lg/dose of CsA (-111-) 91 days, and for the group treated with 10()0 f.Lg/dose (-.-) 85.61 days. Tumours appearcd at almost the same frequency in ali groups tested. Stati­stics (ANOVA One Way) have proved that there is no difference in latency period between the groups tested (p > 0.05). Mice died from 19 to 91 days after tumour detection (Figure 2). The average survival tirne after tumour occurrence for the control group was 37.5, for the group receiving 250 µ,g/dose of CsA 35.27 days and for 1000 µ,g/dose treated Kotnik V and Banic S group 38.84 days. There was no statistically significant difference between the groups tested (p > 0.05). , • • • • • • ro • • ™ Survival tirne after tumour appearance (days) Figure 2. Survival tirne after tumour appearance. Mice died from 19 to 91 days after tumour detection. The average survival tirne after tumour occurrence for the control group (-•-) was 37.5 days, for the group recciving 250 1-Lg/dosc of CsA (-111-) 35.27 days, and for group treated with 1000 1-Lg/dose (-.-) 38.84 days. There was no statistically significant difference bet­ween the groups tested (ANOVA One Way) (p > 0.05). In Figure 3 the tirne from the injection of carcinogen to animals' death is presented. Mice began to die on Day 77, and were ali dead by 0,9 o,, 0,7 0,6 j 0,5 e a.. 0,4 M 140 B - Time to death aftertumour induction {days) Figure 3. Time from the injection of 20-methyl cholan­trene to animals' death. Mice began to die on Day 77 and were ali dead by Day 193, except for one indivi­dual in the control group (-•-) and one individual in the group treated with 1000 /.Lg/dose of CsA (-.-). There were no statistically significant differences bet­ween groups tested (ANOV A One Way) (p > 0.05). Day 193, except for one individual in the con­trol group, and one individual in the group treated with 1000 µ,g/dose of CsA. There were no statistically significant differences between groups (p > 0.05). It is interesting to notice that in the control group one animal survived without tumour and one mouse died intercurently, while in the group treated with 250 µ,g/dose of CsA two mice died intercurrently; in the group treated with 1000 µ,g/dose of CsA one animal survived without turno ur and six died intercurrently ( data not presented). Discussion In the case of CsA, which has no direct geno­toxic effect, tumour promotion is probably dose-dependent.1• 16 This observation led to the idea to test the possible role of CsA in a model of 20-methyl cholantrene induced solid subcuta­neous tumours in mice. 14• 17 It is allready known that CsA inhibits synthesis and excretion of 11-2, but how this function is performed remains unclear. 18 From the pharmacological studies it is known, that CsA binds to Iymphocytes, Ieu­kocytes and to a variety of celi lines in a 19• 20 specific, saturable and reversible way. In whole blood, CsA is mainly associated with erythrocytes which contain the Cyclosporin­binding protein ciclophilin. For celi membrane binding a membrane glycoprotein gp170 is im­portant. It is physiologically expressed in kidney and !iver cells and in multi-drug resistant tu­mour cells.21 CsA in cells inhibits the gene transcription for 11-2 synthesis. The gene pro­moter is the most likely part of the gene affect­ed. In contrast to generally accepted inhibitory effecs on 11-2 synthesis, CsA inhibits the trans­cription of a variety of other activation-induced cytokines and proto-oncogens. It is Iess known that CsA inhibits celi growth of severa! tu­mours. Concentrations in the range of 10 µ,M inhibit tumour celi proliferation.22 These con­centrations may either inhibit the protein syn­ Cyc/osporin treatment and induction of truas thesis or are allready cytotoxic. Those tumours that grow only in the presence of growth factors are the most inhibited by CsA. In conclusion, CsA is not mitogenic and exerts no proliferative effect on any of the observed experimental system. In genotoxicity tests it was shown that it is devoid of any mutagenic properties. Carcinogenicity of CsA was tested in a variety of animals. In mice (OF-1 strain) fed with O, 1, 4 and 16 mg/kg of CsA for 78 weeks, an increased mortality of mice was established only for high-dose treated females.23 Malignant lymphomas were found both in controls and treated animals at a high incidence, without statistically significant effect of the drug treat­ment. 1, 3 . 10 An interesting finding was that CsA may alter the type, frequency and distribution of spontaneously occurring osteomas in OF-1 mice. 1 Mice treated with N-methyl-N-nitrosou­rea had a reduced latency period for the deve­lopment of tumours, however the type and distribution of tumours were not influenced by CsA administration.24• 25 In rats, CsA allowed a faster development of gastrointestinal tu­mours induced by N-methyl-N-nitro-nitrosogua­nidin treatment. 26 There is an interesting report on protective effect of CsA on 3-methylcholan­trene-induced lymphomas by oral administra­tion in the rat.27 In summary, ali the Figure 2. Doublc-barrclled typc carina] reconstruction with right lowcr and lcft stcm bronchus (in combina­tion with right upper bilobcctomy). Carina/ resection and reconstruction double-barrelled type References Trenc/s Gen Thorac Surg. Philadelphia: Saunders, 1987; 2: 91-106. 5. Mathisen DJ, Grillo HC. Laringotracheal trauma. 1. Hetzel MR, Smith SGT. Endoscopie palliation of Acute and chronic injuries. In: Grillo HC, Escha­ trachcobronchial malignances. Thorax 1991; 46: passe H eds. Jnt Trends Gen Thorac Surg. Philadel­325-33. phia: Saunders, 1987; 2: 117-23. 2. Grillo HC. Carina] rcconstruction. Ann Thorac 6. Montgomery WW. Suprahyoid relcase for tracheal Surg 1982; 34: 356-73. anastomosis. Arch Otolaryngol 1974; 99: 255-60. 7. Pearson FG, Todd TRJ, Cooper JD. Experience 3. Mathisen DJ, Grillo HC. Carina] resection for with primary neoplasms of the trachea and carina. bronchogenic carcinoma. J Thorac Cardiovasc Surg J Thorac Cardiovasc Surg 1984; 88: 511-18. 1991; 102: 16-23. 8. Grillo HC, Mathisen DJ. Primary traceal tumors: 4. Perelman MI, Korolcva NS. Primary tumors of the treatment and results. Ann Thorac Surg 1990; 49: trachea. In: Grillo HC, Eschapasse H eds. Int 69-77. Radio/ Oncol 1996; 30: 36--40. Pneumonitis after bronchoplastic surgery in stage I lung cancer Igor V. Kouzmine, Vladimir P. Khartchenko Department of Thoracic Surgery, Moscow Research Institute of Diagnosis and Surgery, Russia Fram the year 1965 to 1990 1959 patients underwent complete resection far primary lung cancer. In 910 cases stage 1 was histologically proved. 827 (90.9%) patients were observed 5 years after treatment, 551 (60.3 %) -10 years. Following bronchoplastic lobectomy without irradiation in 4.5 % cases local recurrence was observed, following this operation combined with irradiation -1.5 %. After lobectomy and radiation therapy (350 cases) remote lesions of the lung were observed in 20 % without any impact of bronchoplasty. Following conservative resection (less than lobectomy) the rate was 12.8 % (p<0.09). Remote lung lesions correlated with the incidence of bronchial obstruction. Survival of patients in complete remission depended on lung parenchyma volume sacrificed. Pneumonitis had not directly influenced survival in remission. Key words: lung neoplasms -surgery; lung diseases, interstitial lntroduction Approximately 70 % of stage I lung cancer patients survive at least 5 years. The problem of quality of their life is growing more and more insistent. So the fundamental principles in modem management (early diagnosis, organ­preserving treatment with multidisciplinar ap­proach and second primary cancer screening) have not been yet thoroughly estimated in pulmonary oncology. 1 Irradiation of the lung produces two histolo­gic phases of damage. Radiation pneumonitis appears within 6 months after treatment. It is Correspondence to: Prof. dr. Igor V. Kouzmine, Mo­skow Research Institute of Diagnosis and Surgery, 117837, GSP-7, Profsojuznaya ul., 86, Moscow, Rus­sia; Fax: +95 334 29-24; E-mail: director@ mniidih. UDC: 616.24-006. 6-089-06;616.24-002 characterized by edema, deposition of fibrin­like material in the alveolar spaces forming hyaline membranes and degeneration of epithe­lial cells lining the alveolar walls specifically type II pneumocytes. Consequently, the predo­minant lesions is a progressive diffuse fibrosis, multiple microathelectases, multifocal oblitera­tion of the air spaces and replacement of the lung parenchyma with scar tissue. Secondary vira! and bacterial infections produce refractory chronic pneumonia. These variety of atypical inflammation can hardly be distinguished from irradiation pneumonitis, the first phase of the complication, especially in combination with surgery. Terminology of late chronic lung le­sions and atypical pneumonia (pne um oni tis) has not been clearly set up.1 For many years we were greatly interested in optimal strategy in the treatment of early lung cancer. In order to prevent the local Pneumonitis after bronchoplastic surgery in stage 1 lung cancer recurrencies after bronchoplastic and conserva­tive resections we used preoperative (PreRT) and postoperative external irradiation (PostRT). The remote iatrogenic complications and concomitant diseases were studied for more than twenty years. Thoracic surgeous, radiologists and cancer practitioners should be made aware of, and familiar with setbacks of organ preserving treat­ment. We will try to share some experience that seems to be important not only for radio­logists, but also for broad specter of medica! specialists. Patients and methods From 1966 to 1990 complete resection was carried out in 1959 cases of Jung cancer. In 910 patients stage I was pathologically proved. Car­cinoma in situ -3 (0.3%), Tl -375 (41.2%), T2 -532 (58.5 % ) cases. The primary lesion localization, histology and sex ratio had no peculiarities. The analysis was based on the patients status at the end of April 1993. Survival was counted according Kaplan­Meier and Fisher methods with the use of computerized data base (FoxPro). 415 patients were alive including 336 in complete remission and 29 with recurrences. From the year 1983 we used prospective randomization for three regimens of manage­ment: surgery, PreRT, PostRT. Methods of treatment are provided in Table l. Table l. Stagc I Jung canccr: numbcr of casc and trcatmcnt rcgimcns; in brackcts -PrcRT+PostRT. Surgcry I r r a d i a t i o n Nonc PrcRT PostRT Sum Conscrvativc Lob-BiJobcctomy Pncumoncctomy 41 277 42 19 (15) 147 (17) 30 (6) 64 (15) 224 (17) 28 (6) 139 665 106 To ta 1 360 196 (38) 316 (38) 910 Central cancer was observed in 335 patients. Lobectomy mainly with bronchial resection un­derwent 223 (66.3 % ), segmentectomy with bronchoplasty 6 patients. Bronchoplastic seg­mentectomy technique with original automatic suture was presented elsewhere. 1• 2 Lobectomy with bronchial resection in central lung cancer was performed in 271 cases. Appro­ximately 80 % of wedge and sleeve resections were connected with right upper lobe bronchus and the least of bronchoplastic operations ­with left lower lobe bronchus. PreRT with betatrone 25 Me V or telecobal­therapy was completed in 550 cases according to 3 protocols: total dose -40-45 Gy (2-2.5 Gy daily fractions); total dose -36 Gy (3 Gy in 12 fractions); single dose 7.5 Gy with thoracotomy on the next day. The irradiated vol ume included hilus, bifurcation and paratracheal zone on the side of the lesion. Lung resection was practiced 3-4 weeks following completion of radiation therapy. PostRT was accomplished in 512 cases. In 38 cases of PreRT (7.5 Gy) added PostRT (30Gy). Result Definite Iocal effect of adjuvant irradiation was observed in patients subjected to bronchoplastic and conservative procedures. Following bron­choplastic lobectomy without irradiation in 99 cases of central stage I lung cancer 5 ( 4.5 % ) local recurrences were observed, after this ope­ration combined with radiation therapy (124 cases) -in 1 (1.5 % ). The difference in the rate of Iocal recurrences is evident in Tl and T2 tumors (Figure 1). NONE ADJUV ANT Figure l. Ratc of Jocal rccurrcnccs following broncho­plastic Jobcctomy with and without adjuvant thcrapy in stagc I Jung canccr. Kouzmine I V and Khartchenko V P But this gain was paid by radiation pneumo­nitis. The rate of remote lesions of the operated Jung was studied after lobectomy as typical resection in stage I peripheral and central Jung cancer. Combined treatment completed in 350 patients, including 140 with wedge or sleeve resection of the bronchus. Following broncho­plastic procedures Iesions of the preserved Jung lobe was found in every tenth patient. Broncho­plasty generally had no impact on the rate of pneumonitis (Table 2). Table 2. lncidence of pneumonitis following lobectomy and bronchoplasty (p>0.05). Remote lesion Lobectomy of the lung No With bronchoplasty bronchoplasty Fibrosis & mild clinical 2.4% 2.1% symptoms Fibrosis & marked 11 % 7.8% clinical symptoms Abscesses 1.4% 2.8% Bronchitis 3.8% 4.3% Bronchiectases 0.9% 2.1% Early postoperative athelectases and pneu­monia had no definitive connection with remote Jung Iesion on the diseased side. In 53 patients with these postoperative complications remote lung Iesions were observed in 9 (15.2 % ) cases. Of the 613 patients that had no complications after lobectomy, pneumonitis was detected in 76 (12.4 % ) cases (p>0.05). There was no significant difference in the incidence and character of pneumonitis after PreRT and PostRT. The rate of Jung Iesions was somewhat higher following 36 Gy total 0.05 Bronchiectases o 2.1 % >0.05 The rate of remote pneumonitis in patients with bronchial passage disturbances was 72.7 % , without any dependence on the type of surgery. In patients that had no complications this rate was 10.5 % (p<0.001). We failed to demonstrate any statistically significant difference in the rate of remote Jung damage in connection with method of bronchial resection, but after wedge lobectomy the pa­tients had some more complications than after sleeve resection. It was influenced by freguency of bronchial curve at the anastomotic area. Survival of patients in complete remission depended on the volume of sacrificed Jung parenchyma (Figure 2). In order to find out the influence of pneumo­nitis on fatal concurrent diseases we considered the causes of noncancerous deaths of stage I lung cancer patients. The results are presented in Figure 3. There were no obvious differences from natural mortality of the general popula­tion. Rate of infectious Jung complications are not of primary importance. Four patients had Pneumonitis after bronchoplastic surgery in stage I lung cancer 100?:----- ---­ 80"':-···········• .•........ · 60.'. ... ...... ····• .. ............. . > ........... o: .tc:c;" => (f) 20.. 12:3°1,5678910 YEARS "' · ConserY.res. ········ Pn-ecton·1:r --Lobecton1T Figure 2. Survival of stagc I lung canccr paticnts in complctc rcmission aftcr conscrvativc surgcry, pncu­moncctomy and lobcctomy. PNEUMONIA {21.0°0}-1 INFARCT & INSULT Figure 3. Noncanccr causcs of dcath aftcr curativc trcatmcnt of stagc I Jung canccr. remote fatal pulmonary hemorrhage following sleeve lobectomy. In three of them there were no signs of pneumonitis or definite endoscopie disturbances of the anastomotic area before initial hemophtysis began. Among patients that had died of heart and lung insufficiency there were some cases of pneumonitis. But in our series survival of the patients with remote pneumonitis was some­what better than in group without this compli­cation (Figure 4). Discussion Most of investigators are in consensus that curativity of bronchoplastic lobectomy is com­ 100.------------­ > o: ::::, (.f) 40;'{ 20::: . 0%------­1 2 3 4 5 6 7 -B ·-·­9 YEARS --,dtl1 pne1-1n,onitis no pneurr101ritis Figure 4. Survival of stagc I lung canccr paticnts with and without pncumonitis in complctc rcmission. parable with that of pneumonectomy, but the quality of Jife is significantly better. Meanwhile the rate of locaJ recurrences after organ-preser­ving surgery is rather high. The average inci­dence of noncancerous stenoses -10 % . 3-4 We have used PreRT and PostRT from the early 1960th.5 Few authors investigated the adverse effects of irradiation in combination with bronchopJastic resections.6• 7 But their ex­perience in stage I Jung cancer management was not enough to make any definite conclu­sions. Both radiation reactions and common pneu­monitis manifest in productive cough and fever. Radiological investigation reveaJs the diffuse densities of operated Jung, multiple subsegmen­taJ athelectases and shadows of inflammation, local fibrosis of pJeura. Than follows the pro­ductive alveolitis, fibrosis of septae and exudate organization. In cases of marked infection the atypicaJ chronic pneumonia, bronchiectases, ab­scesses develop. In these circumstances one can hardly differentiate the adverse effect of irradia­tion and impact of surgicaJ trauma. In most severe cases these patients need repeated hospi­talizations for unspecified treatment. The important role in prevention of postope­rative pneumonitis plays the improving of surgi­cal technique and the use of absorbable sutures in constructing the bronchiaJ anastomosis. Com­pJications in the anastomotic area were reported Kouzmine I V and Khartchenko V P in 12 % of cases.3-4 In contemporary series the rate is to be lower. The explanation of relatively good survival of patients with pneumonitis is difficult to find out. One could presume that in group with this complication were gathered the cases with aug­mented reaction to autoimmune and infectious irritation. It is worth to remind that by chance the OECI conference on cancer and quality of life was held quite near the place where four and a half centuries ago Theophrast von Hugenheim named Paracelsus began his medica! practice. He was the pioneer of prevention and chemot­heraphy of lung cancer. Who knows, maybe his methods of improving performance status were more effective than ours. He lived quite near Bled in Villah, former Corinthia. It is high tirne to suggest to our hospitable Slovenians and dear guests to inspire his positive medieval principles for the management of lung cancer patients even in complete remission. Conclusions l. Adjuvant radiation therapy decreases the probability of local recurrences after broncho­plastic resections in stage I lung cancer. 2. Pneumonitis is the major handicap of or­gan-preserving treatment. Remote lesions of the preserved !obes after combined treatment occur in 10-20 % cases of bronchoplastic resec­tions. 3. Stenoses and deformation in the bronchial anastomotic area happen in 7-15 % of patients. This remote complication do not significantly correlate with the type of bronchoplasty and essentially impact on the rate of pneumonitis. 4. The quality of life of stage I lung cancer patients decreases with the intensity of lung resection. 5. The least morbidity and maximal survival in complete remission of stage I lung cancer patients prove the conservative resections and remote mortality has no direct relation to pneu­monitis. 6. Sudden pulmonary hemorrhage, the typi­cal complication of sleeve resection of the bron­chus, usually bursts our without pneumonitis or obvious endoscopic disorder of the anastomotic area. References l. Khartchenko VP, Kouzmine IV. Lung Cancer. Moscow, 1994. 2. Khartchenko VP, Chkhikvadze VD, Eltyshev NA, Kouzmine IV. Segmenta! Jung resection with sta­pler. Grudnaja i serd-sosud, Khir (Rus) 1993; 6: 57-60. 3. Maeda M, Nakamoto K, Ohta M, et al. Statistical survey of tracheoplasty in Japan. J Thorac Cardio­vasc Surg 1989; 97: 402-14. 4. Kawahara K, Tomita M, Ayabe H, Kimono K. Bronchoplastic procedures with angioplasty for Jung cancer. Acta Med Nagasak 1988; 33 (1-4): 203-7. 5. Khartchenko VP, Volochov BE. Combined treat­ment of lung cancer patients. Matcrials of scientific studies on oncology. Moscow, 1966: "48-52. 6. Paulson DL, Urschcl HC, McNamara JJ, Shaw RR. Bronchoplastic proccdurcs for bronchogenic carcinoma. J Thorac Cardiovasc Surg 1977; 73: 927-47. 7. Fabcr LP, Jcnsik RJ, Kittle CF. Results of slecve lobectomy for bronchogenic carcinoma in 101 pa­tients. Ann Thorac Surg 1984; 37: 270-84. Radio! Oncol 1996; 30: 41-5. Pneumatosis intestini and pneumoperitoneum in acute lymphoblastic leukaemia Marija Frkovic,1 Ante Mandic,1 Nada Jonic,1 Boris Labar,2 Mirando Mrsic2 1 Department of diagnostic and interventional radiology, 2 Division of haematology, Department of medicine, School of Medicine, University of Zagreb, Clinical Hospital center Rebro, Kispatic str no 12, 41000 Zagreb, Croatia Pneumatosis intestini and pneumoperitoneum can originate after infection or gastrointestinal obstruc­tion. Very rarely this condition develops as a random event in patients with collagen disorders and vasculitis, acute and chronic graft versus host disease and various immunodeficiency with no clinical impact. A patient with acute lymphoblastic leukaemia who developed pneumatosis intestini and pneumope­ritoneum with no clinical evidence is reported on. According to the clinical data we assume that the etiology of this condition in the patient is multifactorial: leukaemia, chemotherapy, corticosteroids, sepsis and/or fungal infection. Correlation between patient clinical status and radiological evidence of pneumatosis peritonei is crucial to obtaining the right picture of clinical situation. Key words: leukaemia, pneumoperitoneum; pneumatosis intestinalis Introduction Pneumatosis intestini is a common radiological finding in patients with intestinal necrosis. Inte­stinal perforation is closely accompained with "surgical" pneumoperitoneum and peritonitis. This serious condition has typical clinical finding of acute abdominal pain which leads to urgent abdominal operation. In contrast to this situation "benign" pneuma­tosis1 or "internistical" pneumoperitoneum2 Correspondence to: Marija Frkovic, M.D., Ph.D., Maksimirska street no 12, 41000 Zagreb, Croatia. usually has no clinical signs and can be diagno­sed only by plain abdominal film. This "benign" pneumatosis is caused by other extraintestinal diseases or conditions. Such patients are treated by conservative therapy and very rarely by surgical operation.3• 4 The main purpose of this case report is to point out specific radiologic signs of "benign" pneumatosis and pneumoperitoneum. Case report A forty-seven-year-old male was admitted to the hospital because of weakness, high number of WBC, anemia and a low number of platelets. UDC 616.155.392-036.11-06:616.381-003.219 Bone marrow aspiration showed immature cells Frkovic M et al. compatible with acute lymphoblastic leukaemia. Chemotherapy by daunorubicine, cyclophos­phamide, vincristine and 6-methylprednisolone was introduced (day + 1). Ten days later, mu­cositis grade IV was observed. At the same tirne the patient became febrile and blood cul­ture revealed infection with Streptococcus viri­dans. Antibiotics regimen, consisting of netyl­micin and penicillin, was started. At the same tirne chest X-ray showed multiple infiltrates in both upper Jung fields compatible with fungal infection. Amphotericin B and 5-flucitosine was introduced. Unfortunately, bone marrow aspi­ration on day + 23 showed 90 % of blast, and Figure 1. Pncumoperitoneum: gas shaped like a sicklc salvage chemotherapy of intermediate doses show subdiaphragmatic and subhcpatic. consisting of arabinoside and m-amsacrine over 6 days was started. During ali this period patient did not have signs of abdominal pain or surgical abdomen. On day + 39 routine chest X-ray was perfor­med because of previously diagnosed multiple Jung infiltrates. Examination of this X-ray sho­wed regression of Jung infiltrates but with air shaped like a sickle under both sides of diap­hragm. Patient was subfebrile, 37.5 ° C, but in a good condition without any pain or other symptoms. Karnofsky score was 70 % . Clini­cally we found gentle abdomen with enlarge­ment of !iver (3cm bellow right costal margin). There was an evidence of intestine rnovernent. Plain-filrns of abdomen showed evidence of pneumoperitoneurn, with air in wall of small and large intestine, especially in transverse co­lon and mesocolon (Figure 1 and 2). Air collec­tions, were oval or rounded but in some parts irregular shaped. This air mass parallely follo­wed intestinal contour. In mesocolon, it was radially placed but parallely followed the rne­senterium vessels. There was no evidence of significantly distinct intestine meteorisrn. Patient position had no influence on the above mentioned radiologic finding. Radiologic evidence of pneumoperitoneurn and pneumatosis intestini slowly decreased over 11 days without any clinical impairment. Figure 2. Pneumatosis intestinalis: gas in thc wall of the small and large bowel, cspccially in transverse colon and mesocolon. On day + 50 there was no evidence of air in abdominal cavity. During ali this tirne the pa­tient was maitained on fluid replacement, inte­stinal decontamination, antibiotics, antifungal and sterile diet. Bone marrow aspiration on day + 65 showed presence of blasts. Because of that patient received high doses of cytosine arabinosie over 6 days. This period was complicated with heart fai­lure and patient died on day + 75 after starting of initial chemotherapy course. Pneumatosis intestini and pneumoperitoneum in acute lymplzoblastic leukaemia Discussion Intestinal pneumatosis is collection of air in intestinal wall especially in submucosa and sub­serosa. "Benign intestinal pneumatosis" was first des­cribed in 1973. 1 It is a common finding in newborns with severe necrotic enterocolitis.4-10 There were re­ports on the same condition in patients with leukaemia and lymphoma,1 1-14 with collagen 16 disorders, 15· with severe respiratory distress syndrome,2· 17· 18 in acquired and hereditary im­ 12 • 19- 22 munodeficiency disorders,7• and after 2 1 immunosupressive therapy. 19, Etiologycal explonation of intestinal pneuma­tosis is multifactorial. 5• 23· 24 Conditions such are disease (leukaemia, lym­phoma, malignant diseases, immunodeficiency and collagen disorders) chemo-radio therapy, administration of corticosteroides, bacterial, vi­ra! or fungal infection could cause mucosal damage and intestine wall necrosis. Severe anaemia in patients with malignant disease might cause ischaemic intestine wall 14 necrosis. 5• Administrations of corticosteroides, espe­cially in infancy, might develop atrophia of lymphoid tissue in small and large intestine which leads to submucosal or subserous air disection.12· 21 A patient receiving high doses of chemora­diotherapy, as a condition regimen prior to marrow transplatation, may experience this condition due to the toxic effect on gastrointe­stinal tract. Although, gastrointestinal tract is a target tissue for acute and chronic graft versus host disease. According to Keats 13 ancl Y eager20 mucosal damage appeares 2-3 weeks after finishing che­motherapy. One of the major factor in developing such condition is presence of bacteria, especially anaerobes, in the bowel.5• 6• 23 In some patients with intestinal pneumatosis infections with rotaviruses and aclenoviruses were documented. 25 Vira! enteritis in patients receiving marrow graft is a very rare and there are no availablc literature data on this condition. Clinical features of pneumatosis intestini usually consist of abdominal pain. In some cases, so called "silent" or "indolent" cases pneumatosis intestini ancl pneumopcrito­ncum is a random fincling on routine chcst X-ray_ ll, 12. 24 The main site of this air mass is terminal ileum, large colon, especially colon transversum and ascendent,8• 13· 20 rarely in aboral part of colon.26 In our patient chemotherapy was startcd 39 days before routine chest X-ray were perfor­med. The patient experienced no abdominal sym­ptomatology. Plain abdominal film showed air shaped like a sickle under both sides of the diaphragm in diameter of 1 cm. Pneumatosis was extended to the terminal ileum, mesocolon and large intestine up to the rectosigmoidal border. Finding with such extension as in our patient, 1926 without clinical impairment, is very rare.12 · • Usually, evidence of air in abdominal cavity disappears within two weeks. 13 In our patient 11 days after first X-ray there was no evidence of air collection in the wall of small and large intestine and mesocolon. The characteristics of "benign" asymptomatic pneumatosis are oval or rounded air cystic collections in the serosa and subserosa. This distinguishes "benign" pneumatosis from sym­ptomatic pneumatosis in which air collection is linear and narrowly located in the submuco­sa_ 3. 27 "Benign" pneumatosis is also easily distin­guished from pneumatosis cystoides intestinalis in which air collections accumulate in the large oval cysts like spaces along the intestine wall. Moreover, in "benign" pneumatosis patient's position (standing vs lying) and tiJne (24 hour period between X-rays) have no influence on the position of air in abdominal cavity. 8 In our opinion, such differentiation of the intestinal pneumatosis, despite the fact that it emphasises the importance of radiological fin­dings obtained by standard examinations, which Fr kovic M et al. is the aim of this paper, is not accurate unless clinical symptomatology of a patient is not build in the diagnosis as a whole. Similar attitudes are also presented by Tho­mas VI.28 Some recent studies report usefulness of ul­trasound and computerized tomography in de­termining location and progression of intestinal 27 31 pneumatosis,9• -, but the main diagnostic method is stili plain abdominal film. In conclusion, we agree with studies of Keats13 and Yeager20 who emphasize that a well informed radiologist and carefully exami­ned series of plain abdominal films may give a very useful information on the development of pneumatosis intestini and "benign" pneumope­ritoneum. Correlation between patient's clinical status and radiological evidence of pneumatosis inte­stini and pneumoperitoneum is crucial in obtai­ning the right diagnosis. References l. Me Swceny WJ and Stempel DA. Noniatrogenic pneumomediastinum in infancy and childhood. Pediatr Radio/ 1973; 1: 139--44. 2. Zerella JT, Me Cullough JY. Pneumoperitoneum in infants without gastrointestinal perforation. Sur­gery 1981; 89: 163-7. 3. Knechtlc SM, Davidoff AM, Rice RP. Pneumato­sis intestinalis: Surgical menagement and clinical outcomc. Ann Surg 1990; 212: 160-5. 4. Sibbald WJ, Sweeney JP, Inwood MJ. Portal venous gas (PVG) as an indication for hepariniza­tion. Am I Surg 1972; 124: 690-3. 5. Sidney WN. Ekstraluminal gas coleetions due to diseases of the gastrointestinal tract. Al R 1972; 115: 225-48. 6. Chabot VH, Slovis ThL and Cullen M. Reeurrent pneumatosis intestinalis in young infants. Pediatr Radio/ 1992; 22: 120-2. 7. Sivit CJ, Josephs ShH, Taylor GA and Kushner DC. Pneumatosis intestinalis in children with AIDS. AIR 1990; 155: 133-4. 8. Rosenquist CJ. An unusual pattern of intramural gas in small bowel infarction. Radiology 1971; 99: 337-8. 9. Vijayaraghavan SB. Sonographic features of pneu­matosis of the smaU bowel. I Ciin Ultrasound 1990; 18: 579-81. 10. Dudgeon DL, Coran AG, Lauppe FA, Hodgman JE and Rosenkrantz JG. Surgical management of acute necrotizing cnterocolitis in infancy. I Pediatr Surg 1973; 8: 607-14. 11. Gala! O, Osse G, Wcigel W, Gahr M. Pneumato­sis intestinalis in children with leukaemia: report of three cases. Eur I Pediatr 1981; 137: 91-3. 12. Borns PF, Johnston TA. Indolent pneumatosis of the bowel wall associated with immune suppressive therapy. Ann Radio/ 1973; 16: 163-5. 13. Keats TE, Smith TH. Benign pneumatosis intesti­nalis in childhood lcukaemia. AJR 1974; 122: 150-2. 14. O'Conncll DJ and Thompson AJ. Pneumatosis coli in non-Hodgkins lymphoma. BIR 1978; 51: 203-5. 15. Muellcr ChF, Morchcad R, Michcner W. Pneuma­tosis intestinalis in collagen disorders. AJR 1972; 115: 300-5. 16. Wang CL, Wang F, Wong KC, Jeyamalar R. Benign persistent pneumoperitoneum in systemic sclerosis. Singapore MJ 1993; 34 (6): 563-4. 17. Kiryu T, Kobayashi H, Kawaguchi S, Kanou S, Uwabe Y, Sakai M, Matsuoka T, Nagata N. A case of status asthmatieus eomplicated by with pneumoperitoneum during meehanical ventilation therapy. Japanese lournal of Thoracic Diseases. 1993; 31 (6): 771-4. 18. Kandylakis S, Makrigiannakis A, Mirra N, Perpy­rakis G, Kontakis K, Apalakis A. Pneumoperito­neum without perforation. Minerva Chirurgica 1993; 48 (12): 717-20. 19. King S, Shuckett B. Sonographic diagnosis of portal venous gas in two pediatric !iver transpalant patients with benign pneumatosis intestinalis. Pe­diatr Radio/ 1992; 22: 577-8. 20. Yeager AM, Kanof ME, Kramer SS, Jones B, Sara] R, Lake AM and Santos GW. Pneumatosis intestinalis in children after allogenic bone marrow transplantation. Pediatr Radio/ 1987; 17: 18-22. 21. Day LD, Ramsay KCN and Letourneau JG. Pneu­matosis intestinalis after bone marrow transplanta­tion. AIR 1988; 151: 85-7. 22. Lagundoye SB and Itayemi SO. Tension pneumo­peritoneum. Br I Surg 1970; 57: 576-80. 23. Yale CE, Balish E, Wu JP. The bacterial etiology of pneumatosis cystoides intestinalis. Arch Surg 1974; 109: 89-94. 24. Pfister J, Riedtmann-Klee HJ. Idiopathic sponta­neous pneumoperitoneum. Case discussion based on two cases, asscssment procedure and therapy and review of the literature. Helvetica Chirurgica Acta 1993; 60 (1-2): 49-56. Pneumatosis intestini and pneumoperitonewn in acute lymphoblastic /eukaemia 25. Capitanio MA and Greenberg SB. Pneumatosis intestinalis in two infants with rotavirus gastroen­teritis. Pediatr Radio/ 1991; 21: 361-2. 26. Chippindale AJ and Desai S. Two unusual cases of pneumatosis coli. Ciin Radiol 1991; 43: 180-2. 27. Candill JL, Rose BS. The role of computed tomo­graphy in the evaluation of pneumatosis intestina­lis. J Ciin Castroenterol 1987; 9: 223-6. 28. Thomas VIL, Cohen AJ, Omiya B, McKenzie R and Tominaga G. Pneumatosis intestinalis associa­ted with needlc catheter jejunostomy tubes: CT findings and implications. J Comput Assist Tomogr 1992; 16: 418-9. 29. Connor R, Jones B, Fishman EK, Sigelman SS. Pneumatosis intestinalis: role of computed tomo­graphy in diagnosis and management. J Comput Assist Tomogr 1984; 8: 269-71. 30. Kelvin FM, Korobkin M, Rauch RF, Rice RP, Silverman PM. Computed tomography of pneuma­tosis intestinalis. J Comput Assist Tomogr 1984; 8: 276-8. 31. Chadba D. Kedar RP, Malde HM. Sonographic detection of pncumopcritoneum: an expcrimental and clinical study. Australasian Radiology 1993; 37 (2): 182-5. Radio[ Oncol 1996; 30: 46-54. Causes of f ertile disturbances in oncological male patients Viljem Kovac Institute of Oncology, Ljubljana, Slovenia It has been known for a long tirne that oncological therapy may influence male fertility while sometimes neglected but important cause of infertility in cancer patients is the fact that certain malignancies exert an adverse effect on the testis before treatment. Therapeutic cancer regimens may have significant, undesirable consequences on the fertility. Surgical treatment can damage the neurovascular mechanisms controlling erection, emission and ejaculation what can prevent the delivery of spermatozoa and result in subfertility. Radiation therapy is used in a number of malignancies including Hodgkin's disease and germ cel! tumours and disturbances in gonadal function through various mechanisms. A majority of men treated with chemotherapy are affected with indeterminate periods of azaospennia, decreased libido and erectile dy:,function. Additionally, we have to consider in oncological patients also factors, which can influence on fertility, such as impact of another drugs, endocrine diseases, diseases of external genitals, congenital and heredital diseases, infections and immunological diseases, impact of nutrition, bad habit, environment and psychological factors. Key words: infertility male; neoplasms -therapy; radiotherapy -adverse effects; antineoplastic agents -adverse effects; surgery -adverse effects Introduction Prevalence of infertility varies from country to country. It is estimated that 5 to 8 % of couples have problems with infertility in developed countries but in developing countries up to 30 % of couples have these problems. 1 Refer­ring to the information of the World Health Organization about one in every ten couples wishing to have a child experiences some form of infertility problem. Extrapolated to the Correspondence to: Viljem Kovac, M.D., Institute of Oncology, Zaloška 2, 61105 Ljubljana, Slovenia; pho­ne: + 386 61 1320 068; fax: + 386 61 1314 180. globa! population this means that up to 80 million people may be suffering from infertili­ty. 2 Due to the increase of inflammations of the genital tract, particular caused by sexually transmitted diseases, due to endometriosis, birth postponement of the first child and more stress the number of infertile couples is in­creasing. 2· 3 Although we do not have exact data about the prevalence of infertility in Slovenia, we can, as to the above mentioned information and as to the fact that 10 000 couples marry every year 4 that it exists 25 % out of married relations, conclude that more than 1000 couples confront with the problem of fertility every year.3 So, UDC: 615.277.3.06:616.697 we must be greatly aware of the problem of Fertile disturbances infertility in Slovenia and give a lot of thought to it. The causcs of infertility are multiple. By the average population its cause is in 30 to 50 % in man5 or it is estimated that in developcd coun­trics in 25 to 30 % the cause of infertility is in 6 man whereas in 20 to 24 % in both partners.2· Semen factor is the cause of infertility in 18 % to 31 % , in 15 to 28 % of cases there are many factors which cause the infertility;7 one of these is also the oncological treatment. At the Institute of Oncology 6551 patients were treated in 1991, 3324 of them were male patients. At the age when fertility is the most present (20 to 60 years old), there were L145 paticnts i.e. 34,4 % of ali oncological male paticnts; among thosc there are also such whom the trouble in fertility prcsents a great pro­blem. 8 In other countrics the situation is com­parable to ours. In the Unitecl Statecl there are more than 26 000 men between 17 ancl 50 ycars olcl, who are affictecl with cancer each year.9 The surgical, radiation ancl chcmotherapeutic regimens that are used to attain the present high survival rate of oncological patients often incluce irreversible clamage to the testis. 10-12 However, the modern tberapeutic approacbes to oncological patients are not only aimecl at cure but also ensuring the least possible sidc effects ancl the optimal quality of life wbich naturally includes preservecl fertility.12· 13 This issue bas become particularly important in younger patients with goocl response to therapy, such as are the patients with testicular tumours, Hoclgkin's cliseasc and with turnours of child­hoocl. 1+-17 As to the information of the Regis­ter of Cancer Patients in Slovenia there is a rclatively srnall number of these patients.8 Al­though it is not clear whethcr the incidence of infertility in men is incrcasing clue to environ­ment factors or whcthcr it is better discoverecl, 18 it is founcl out that the number of patients with testicular tumours, Hoclgkin's cliscase and with tumours of childhood is too small to influence essentially on the incidcnce of infertility in a certain nation. The infertility represents a great individual problem in these patients, so doctors should not neglect it. Nonspecific effects of neoplasia on testis famction Sometimcs negleeted but important cause of infertility in cancer patients is the fact that certain malignancies exert an adverse effect on the testis before treatment. 19-21 Hodgkin's di­sease and testicular germ cell malignaneics have associated testieular injury manifested by da­mage to the spermatogenic stem cell line early in the course of the disease not related to local tumour effect or metastasis.22· 23 It is supposed that the reason may be a 25 primary dysgenesis in the remaining testis,24· elevation of serum's HCG hormone in hormo­nally active tumours,21 local effect of the tu­mours on contralateral testis -with temperature in scrotum ancl with vascularisation change,26 or possibly thc influence of stress in young men who suffer from a newly diagnosed malignant disease. 25 Some simply talk about the histological pecu­liarity of patients with testis cancer as many are discovered at the assessment of infertility or about the fact that testis tumours should have occurred more often in patients with common 27 disturbance of spermatogenesis,21, whereas the 11011-specific effect of malignant turnour 011 contralateral testis should have been the most important additional factor influenced 011 sper­matogenesis. It is interesting that other malignant diseases do not badly effect on spermatogenesis at the initial stadia of the diseases. 15 On the contrary, ali malignant tumours in progressed stadia of the disease connected with the patient's cache­xia and bad condition have weak effect on gonadal function. 22 Surgical treatment Surgical treatment of cancer has no direct de\e­terious effect on testicular function. Retroperito­neal lymph node dissection, abdominal aneu­rysm repair, abdominoperineal resection, and operations that damage the neurovascular me­chanisms controlling erection, emission and eja­ 48 Kovac V culation can prevent the delivery of spermato­zoa and res uit in subfertility .19 Retroperitoneal lymph node dissection After a radical retroperitoneal lymph node dis­section, such as perfonned in nonseminomatous cancer and Hodgkin's disease, patients are in­fertile initially as a result of sympathetic nerve damage that interferes with emission and ante­grade ejaculation. 28 The quantity of semen emis­sion can be essentially reduced,29 the damage of lumbar sympathetic ganglia also produces retro­grade ejaculation,30 or the absence of ejaculation can be estimated. 6· 31 Another pelvic operation With the interruption of parasympathetic inner­vation by a pelvic operation -such as cystecto­my, to tal prostatectomy or rectal operation ­ 32 the loss of erection can be expected. 19• It is not so uncommon that abdominal surgery damages the neck of the bladder and causes retrograde ejaculation. 6 This occurs following the disruption of interna! sphincter. Various extensive operations Extensive operations, frequent in the oncologi­cal surgery, may cause with non-specific effect temporary disturbances both in the function of Leydig's cells and in spermatogenesis.33 Vascular injury following hernioplasty Vascular injury and testicular atrophy following hernioplasty mostly occur in surgery of infants and cause fertility disturbances.30 Patients with testicular tumours usually suffer more from ingvinal hernia tl1an others;6 therefore it is clear that with such patients hernioplastic operations are more frequent and consequently the danger of vascular injuries is much greater. Radiotherapy This effect of ionizing radiation to the male reproductive system both by animals and human beings was quite often docu­ mented. 11-13, 15, 16, 19, 34 Radiotherapy of the hypothalamic and pituitary region In oncology the mentioned region is irradiated in case of various pituitary gland tumours, brain tumours, craniopharingeomas, nasopharyngeal carcinomas as well as leukemias and medulobla­stomas. In such situation the hypothalamus or the pituitary gland may be situated in or at the margin of the radiation field. The estimated tumour dose 40 Gy to the hypothalamus causes in 30% deficiencies in gona­dotropin secretion due to the defect of the release of gonadotropin releasing hormone. 35 At dose 40 to 70 Gy already 60 % of patients are with disturbances; at dose up to 24 Gy thes disturban­ces were not noticed. However, an exact correla­tion between dose and type or extent of hypotha­lamic or pituitary hormone deficiencies conse­quently the fe1iility as well -has not yet been described.12• 35 Direct exposure to ionizing rays at total body irradiation Tota! body i1ndiation prior to bone manow transplantation can cause severe oligo-or azoo­spermia but the testosterone levels in the serum were normal. There was no evidence of direct damage to the hypothalamic-pituitary axis. By the total body irradiation usually 10 to 13,2 Gy in 5 or 6 fractions are applied. By the majority of children the gonadal dysfunction is observed at the just mentioned dose. Therefore the long­ 37 term endocrine management is needed.36• Radiation of the infradiaphragmal region By the irradiation of the infradiaphragmal re­gion gonads are irradiated most frequently ­either by means of direct irradiation (which is not often) or by scatter radiation. Due to scatter radiation i.e. by the irradiation of patients with testis tumours azoospermia or oligoastenospermia turns up but such state can later improve.38• 39 lrradiation effects on spermatogenesis with single dose irradiation Single gonadal dose of 0,5 to 0.78 Gy causes marked oligospermia and decreased sperm con­ Fertile disturbances centration in the semen to about two million per ml 11 weeks post irradiation. Spermatids seemed to be damage after 4.0 to 6.0 Gy, since the number of agile spermatozoa was signifi cantly decreased afterwards. 12· 40 However, sin­gle 0.05). The influence of the last quarter 10 20 30 40 50 ľ 49/198 -Newmoon on SP occurrence was more pronounced in women than in men, but the difference was statistically insignificant (p > 0.05) (Figure 2). The relationship between the patient's age and the moon phase at the tirne of SP onset is presented in Figure 3. Patients under 30 years of age were significantly more susceptible to the effect of the moon phase (p < 0.01) than older patients, in whom this effect was statisti­cally insignificant (p > 0.05) . O 10 20 30 40 50 % .. 23/105 33/139 New moon • First quarter 27/105 35/139 f) Full moon l'lllllllllllil 1111os* 27/139 44/10 (t Last quarter 44/139 5+ . expected frequence mlll<30years p < 0.01 * c:::J > 30 years + p < o.os Figure 2. Phases of the moon and age of patients with SP. 10 20 30 40 50 % §..55/244 Newmoon 1 • 51/198 t) F;,st qua,te, l 13/46 64/244 Firstquarter 19/41 f)55/203 a-6141 9/198* Full moon 9/46 . _,J 38/244 * O . O 69/198 * ct Lasi quade, (t Last quarter 19/41 68/203 + 111118/46 87/244* t expected frequence expected frequence Rmll First SP llilllll'il!ii Aecurrenl SP c::J Total * p < 0.01 11111111111111 Women ll!ilill!II Men + p > o.os Figure l. Phases of the moon and number of SP Figure 3. Sex of paticnts and rclative number of SP events. at cach moon phase. The role of phases of the moon in the development of .1pontaneous pneumothorax Discussion References To our knowledge, an associat10n between a specific moon phase and the development of SP has not been described. In fact, there is no evidence to support involvement of the moon in any physiologic or pathologic state in man, although according to popular belief, it affects all human conditions. The moon with its phases may, however, exert an influence on man by gravitation, as it affects the tide. The gravitatio­nal force in the erect human posture has been described as a potential factor precipitating the development and rupture of subpleural blebs and bullae in the upper Jung !obes close to the 7 8 apex. • Also the development of catamenial SP is unclear.9 The temporal distribution of SP events observed in our study suggets that women are more susceptible to the influence of the moon than men. Considering that also the menstrual cycle and the duration of pregnancy in woman are related to the lunar month, our present findings may open a new perspective on the problem of catamenial SP. l. Lichtcr I, Gwyannc JF. Spontancous pncumotho­rax in young subjccts. A clinical and pathological study. Thorax 1971; 26: 409-17. 2. Ohata M, Suzuki H. Pathogcncsis of spontancous pncumothorax with spccial reference to thc ultra­structure of cmphyscmatous bullac. Chest 1980; 77: 771-6. 3. Mclton LJ, Hcppcr NGG, Offord KP. Influence of height on thc risk of spontancous pncumothorax. Mayo Clin Proc 1981; 56: 678-882. 4. Bense L. Spontancous pncumothorax rclatcd to falls in atmosphcric prcssurc. Eur J Respir Dis 1984; 65: 544-6. 5. Garcia CJA, Hcrnandcz CMA, Rego FG, Bustillo FE. Association bctwccn falls in atmosphcric prcs­surc and spontancus pncumothorax (lcttcr). Eur J Respir Dis 1985; 66: 230. 6. Scott GC, Berger R, Mckcan HE. Thc role of atmosphcric prcssurc variation in thc dcvclopmcnt of spontancous pncumothoraccs. Am Rev Respir Dis 1989; 139: 659-62. 7. Kawakami Y, Iric T, Kamishiama K. Staturc, lung height, and spontancous pncumothorax. Respira­tion 1982; 43: 35-40. 8. Fishman AP. Thc normal pulmonaly circulation. In: Fishman AP, cel. Pulmonary disease and dis­orders. New York: Hill McGrcw, 1988: 975-96. 9. Lillington GA, Mitchcll SP, Wood GA. Catamcnial pncumothorax. JAMA 1972; 219: 1328-30. Radio! Oncol 1996; 30: 58-65. A simplified technique for aligning radiation fields in portal imaging Hui Wang and B. Gino Fallone Medica! Physics Unit and Department of Physics, McGill University, Montreal General Hospital, Montreal, Quebec, Canada A simplification of the chamfer matching technique is proposed far aligning radiation fields in portal imaging. In this application, the shaped treatment field is first aligned to the prescribed field by using low order geometric moments. Then the alignment is fine tuned with a simplified chamfer matching technique in which the cost function is successively minimized along coordinate directions. The viability of this minimization approach far the detection of radiation beam shaping errors in portal imaging is demonstrated through comparison with the downhill simplex method. When used in combination with lower geometric moments, this minimization method appears to offer advantages over the standard downhill simplex method in terms of precision and computation speed. Effects of two types of cost functions and edge distance maps in this application are also discussed. Key words: portal imaging; tomography, x-ray computed-methods; radiotherapy Introduction With the ongoing improvement in treatment planning accuracy, there is greater demand on the accurate implementation of a treatment plan. A significant amount of research has been carried out on every aspect in portal image 7 verification, from image acquisition systems1 ­ to post processing algorithmsS-10 and registra­16 tion procedures.11 -Current progress in this Correspondence to: Prof. B. Gino Fallone, Ph. D., FCCPM, ABMP, Medica! Physics Unit and Depart­ment of Physics, McGill University, Montreal General Hospital, 1650 Cedar Avenue, Montreal, Quebec, Canada H3G 1A4; Phone: + l 514 934 8052; Fax: 1 514 934 8229. UDC: 616.149.66-073.756.8 relatively new field and its impact on clinical practice of radiation therapy have been syste­ 17• 18 matically reviewed. To set up a radiation therapy treatment, one needs to register portal images to a reference image in order to visualise the coverage of the target by the radiation beam. Because of the radiation required for portal image acquisition, it is not feasible to perform beam shaping and localization at the same tirne. The common practice is to shape the radiation beam by following the prescription before setting the patient up, then to direct the shaped beam to the desired target by positioning the patient. Accurate beam shaping is not only a factor determining the precision of beam coverage but is also beneficial for the interactive beam loca­lization in this two step procedure. Extracting A simplified technique [or aligning radiation fields in portal imaging anatomical landmarks for image registration is tirne consuming. On the other hand, the radia­tion field is a prominent feature in a portal image. A correctly shaped field can be automa­tically extracted and used as a registration land­mark. Computer programs that can automati­cally align the portal and reference image to the field and show the relative position of the target in the field can serve as a tool for beam localization with an on-line imager. In this paper, we propose a simplified method for aligning radiation fields. A natura! way to detect shaping errors is to align the prescribed field with the treatment field, and to evaluate the visual match. Among ali the techniques that have been employed in this task, chamfer matching is probably the most successful one due to its insensitivity to noise and to discrepancies in shapes. t9 A techni­cal issue in chamfer matching is the minimiza­tion of a cost function of the geometric transfor­mation parameters (translation offsets, rotation angle and scaling factor). Since there is no analytical expression for a cost function, the minimization has to rely on iterative searching techniques among which the downhill simplex 21 method20• and the Powell's method21 have been applied. In a recent study on multileaf collimator configuration verification, Zhou and Verhey reported that the downhill simplex met­hod is not very sensitive to rotation and requires starting points close to the globa) minimum.22 To overcome these shortcomings, they used Hough transform and geometric properties of two contours to determine the starting point of chamfer matching. The downhill simplex met­hod is straightforward and easy to implement but not very efficient in terms of the number of function evaluations that it requires. Powell's method is almost surely faster in ali likely applications.21 If chamfer matching is used in combination with lower order geometric mo­ments, the minimization process will start very close to the global minimum.23 Because of this, standard minimization techniques which are de­signed for general purposes could be simplified to suit specific applications. In this paper, we propose a simplified adaptation of the chamfer technique for matching treatment and prescrip­tion field pairs. Materials and method Thirty pairs of simulation and double-exposure portal films randomly selected from our patient archives were digitized to 512 x 512 digital ima­ges with 8 bit contrast. The prescribed field contours were drawn with a mouse by following the prescription on the simulation images. The treatment field contours were obtained by using a "contour" operation on the field masks auto­matically extracted from the portal images. 1 5 The "contour" operation peels the mask ( an object in a binary image) at the depth of one pixel with an "erosion" operation and then subtract the eroded mask from the original one to acquire the field contour. Edge distance map generation An edge distance map image E(i,j) (Figure le) was generated from a binary image containing a prescribed field contour (Figure la). The value of a pixel in the edge map image is the distance from that pixel to the closest feature (contour) pixel. The higher the value of a pixel, the farther away it is from the contour points. This edge map image, resembling a landscape model with a valley along the prescribed field contour, will be used as a mould onto which the treatment field contour (Figure lb) will be fit. By analogy, the treatment field contour will have different gravitational potential energy at different locations and with different orienta­tions (Figure ld). Registration will be achieved when the treatment field contour slides down into the valley under the action of gravity. Ideally, the value of a pixel in an edge map image should be the Euclidean distance from that pixel to the closest contour point. However calculating Euclidean distances is computatio­nally intensive and yet may not be worthwhile because of digitization effect on the contours. The common approach to generate an edge map is to use distance transformation which approximates globa! distances by propagating Wang H and Fallone B G Figure l. Illustration of procedure for chamfcr match­ing: (a) Simulator image; (b) Portal image; (c) Prescri­bed field contour obtained from (a); (d) Treatment field contour extracted from (a); (e) Edge distancemap generated from (c); (f) After (d) is transformed with a tria! set of parameters, it is overlaid on top of(e). local distances. 24 By analogy, local distance propagation is similar to using a ruler of unit length for measuring long distances. As the ruler is being passed in steps, the number of passes is counted and the total number of counts is used as the distance from the starting point. A distance transformation passes a kernel which carries distances between a pixel and its neighbors over an image and assigns the accu­mulated distances to the pixels it passes by. Our edge distance maps were generated with the "chessboard" distance transformation25 which employes the 3 x 3 kernel 1 1 1 [1 O 1 ] (1) 1 1 1 for representing the local distances and propa­gates the local distances in two steps. In the first step, the upper-left half of the "chess­board" kernel 1] (2) o [. 1 was passed over the contour image (in which ali the contour pixels were assigned the value of O and ali the non-contour pixels were as­signed the value of co) from left to right and from top to bottom. At each pixel, this half kernel was added to that pixel and its four neighbors and the minimum value among the five was assigned to the corresponding pixel in an intermediate image. In the second step, the lower-right half of the "chessboard" kernel o (3) 1 [1 .] was passed over the intermediate image from right to left and from bottom to top. The minimum value among the five was assigned to the corresponding pixel in the distance image. Cost jimction minimization After the edge map (Figure le) was generated from the prescribed field contour (Figure la), a geometric transformation with a set of tria! parameters, where (a, b) is the translation vec­tor, 0 is the rotation angle around the center of mass of the prescribed field, (X„ Y ), and 1p m is the scaling factor, was applied to the treatment field contour { (xi, y); i= 1, 2, ... N} (Figure lb). The transformed contour {(x;, y1); i = 1, 2, ... N} where .r: - , y p cos0 -msinO] [·'• Xp] + [a] (4) 1 [1 1 -sin O m cos O !Ji r' b y. }'.) p was overlaid on the edge map (Figure ld), and the sum of the values of the pixels in the edge map along the transformed treatment field con­tour was used as the cost function N (5) F(a,h,0,m) = LE(x'..y:). i=I F (a, b, 0, m) is a function of the transformation parameters and has its minimum value when A simplified technique far a!igning radiation fields in por!al imaging the correct transformation parameters relating treatment field contour to the prescribed one are used. Minimization of this cost function is usually achieved with iterative searching algo­rithms which start with a set of the initial tria! ) parameters (a0 , b0, B0, m0, navigate in the 4 dimensional space formed by ali the possible transformation parameters, and check some cri­terion at each step until convergence is reached. We used a method similar to that used by Zhou and Verhey to obtain the starting point. Except for the rotation angle (which is set as O), the initial tria! transformation parameters are obtained from the center of mass and the area of the fields: (ao.hu.Oo.mo) (-\'p .'X:t, \-. -\-i.O. /Ii-;) (6) where (Xp, Y p) and Ap are the center of mass and area of the prescribed field, (X,, Y1) and A1 are those of the treatment field. B0 is set to O because the angle between the two contours is always very small before matching in our case. After this preliminary matching, the starting point is very close to the globa! minimum of the cost function. We minimize the cost fun­ction with respect to one variable at a tirne, one after another and cycle after cycle. One dimensional minimization is achieved by brac­keting (Appendix). The cost function is minimized with respect to B first. The initial bracket is set to ( B0, B0 + 0.1). The bracketing stops when F{ao,bo,tJ,., m0)-F(a0,b0,tJ,,+i-rno)/ < 10-·I F(ao.bo.0,,.mo) ' ­ where n is the number of iteration. Similar procedures are then carried out to a, b and 111 sequentially with initial brackets (a0, a0 + 5), (bo, b0 + 5) and (m0 , m0 + 5), respectively. All the procedures are stopped at the same preci­sion. Once a bracketing procedure stops, the corresponding variable is kept at the convergent value. Since our stopping criterion is set very low, one cycle is sufficient to reach convergence for ali the variables. Results and discussion Chamfer matching is a technique of pattern recognition type. The goodness of a match is characterized by a similarity measure. The mea­sure we used is the average edge distance which is the minimum cost. It is defined as the average pixel value along the registered treatment field contour in the edge map image, which can be expressed as k F (am, bm, Bm, 111111), where N is the number of pixels along the treatment field contour and (a,m b11 1, B11„ m111 ) is the fina! trans­formation parameters reached by the minimiza­tion. If the two feature being matched are perfectly similar to each other, thc minimum cost should be zero. Otherwise, it will take on a positive value. When matching simple features like closed field contours, unless the shaping errors are extremely large, a smaller minimum cost should correspond to a better match. Mini111ization approach To test the viability of successive minimizations along coordinate directions in chamfer matching when combined with geometric moments, we compared the average edge distance obtained with !?_Uccessive @inimization .long .oordinate _g_irections, Ds.M.A.C.D. to that obtained with the downhill simplex method Dsimplex· A flowchart of the comparison test is shown in Figure 2. Figure 2. Flowchart of thc comparison test bctwccn thc simplcx mcthod and succcssivc minimization along coordinatc dircctions (S.M.A.C.D). Wang H and Fallone B G o .o § 4 z -64 -56 -48 -40 -32 -24 -16 -8 O S 16 24 32 40 48 56 64 Percentage Difference Figure 3. Comparison of fina! average edge distances achieved with different approaches. (a) successive line minimization versus the simplex method with "chess­board" edge distance maps and arithmetie average edge distance as cost function. (b) arithmctic vcrsus root of mean squarc average edge distance as cost function when "chessboard" edge maps and successive line minimization were used. (c) "chessboard" versus "5-7-11" cdge maps when arithmetic averagc edgc distance as cost function was minimized successive line minimization. Computation speed The two algorithms are also compared in term of computation speed. Table l lists the numbers of iterations to reach convergence. Conver­gence is defined in the following manner: for successive line minimization, either the preci­sion criterion is satisfied or the values of the cost function at the two ends of the bracket no longer change; for simplex algorithm, either the precision criterion is satisfied or the values of the cost function at the vertices of the simplex no long change. It should be noticed Throughout the comparison tests, both minimi­zation procedures are started after the first step match obtained with geometric moments in Eq. ( 6). The vertices of the initial simplex are set to (a0, b0, B0, m0), (a0 + 5, b0, B0, m0), (ao, bo + 5, Bo, mo), (ao, bo, Bo + 0.1, mo) and (ao, bo, Bo, mo + 0.1), where a0, b0, Bo and m0 are given in Eq. (6). The absolute values of the minimum costs obtained with the two methods are very close. In order to see the difference clearly, we calculated the relative difference in percentage x lilll. (7) A VERAGE EDGE DISTANCE 16 avg.=-9.215 % (a) S.M.A.C.D. . !2 vs 0 Sirnplex 'o 8 6 4 i -64 -56 -48 -40 -32 -24 -16 -S O 8 16 24 32 40 48 56 64 32 (b) Arithmetic avg.=-2.271 % vs . 24 RMS 0 15 16 o .o § 8 z -64 -56 -48 -40 -32 -24 -16 -8 O 8 16 24 32 40 48 56 64 16 avg.=-5.640 % (C) "Chcssboard" . 12 VS 0 "5-7-11" a histogram of which obtained from the 30 cases is plotted in Figure 3a. It can be seen that, on the average, successive line minimiza­tion along coordinate directions can achieve a smaller residual value of the cost function therefore has better precision than the simplex algorithm in this matching scheme. Successive minimization along coordinate di­rections is actually the first step of the Powell's method.26 Starting from the coordinate direc­tion set, Powell's method successively mini­mizes a function in each direction in the set and adjusts the direction set after each cycle of line minimization through ali the variables. The adjustment of searching directions is to handle functions which can not be well approximated by quadratic forms because Powell's method is based on Taylor expansion. For example, a function having a long narrow valley will make successive minimization along coordinate di­rections very inefficient. However, when simple geometric objects like the field contours, are being matched, it is very unlikely that the cost function will have such erratic behavior. More­over, after preliminary matching has been done with geometric moments, the starting point is very close to the globa! minimum of the cost function. In this neighborhood, the cost funct­ion can be well approximated by a quadratic form which can be exactly minimized by one pass of line minimization through all the vari­ables. A simplzfied technique far aligning radiation fields in portal imaging Table l. Numbers of iterations for reaching conver-Effects of cost functions gence in successive line minimization along eoordinate According to the investigation by Borgefors,27 directions (Ns.M.A.C.D.) and the downhill simplex met- the cost function based on the root of mean hod (Nsimplcx)-The average difference (Ns.M.A.C.D. ­ -3.2 square average edge distance (summing up the Nsimplcx) • square of the pixel value along the contour that case number Ns.M.A.C.D. N_\'implcx Ns.M.A.C.D.-N..simplex is being matched when it is fitted into the edge 1 32 31 1 map) has fewer local minima than does the cost 2 31 24 7 function based on the arithmetic average edge 3 33 41 -8 distance. It can, therefore, reduce the chance 4 32 40 5 25 24 1 for minimization being trapped by false conver­ -8 gence. Local minima are not a problem within 6 30 38 -12 our approach since they are bypassed by the -13 8 first approximation [Eq. ( 6)]. To investigate 9 24 33 -9 the effect of the two types of cost function in 10 32 31 1 l1 29 44 -15 this application, we calculated the average edge 12 29 19 10 distance, and observed a slight advantage of 13 34 34 o the arithmetic average type cost function over 14 32 -3 the one of root of mean square average type 42 3 (Figure 2b). In chamfer matching every point 16 28 17 36 44 -8 on the treatment field contour contributes to -1 18 36 the cost function. The root of mean square 19 32 average type cost function increases faster 20 30 32 -2 21 33 33 o around the minimum and therefore assigns more weight to the distorted parts of the treat­ 22 23 33 -10 23 29 32 ment field contour making the minimum of the 24 26 23 3 cost function not correspond as well to the -6 25 31 -6 actual match as does the arithmetic average 26 30 31 -1 type cost fu.ction. 26 26 o Effects of distance transf ormations 29 2 29 31 The accuracy of chamfer matching also depends -1 30 31 that not only less number of iterations are required for sequential bracketing (3.2 on the average), but the number of calculations in­volved in a single iteration in bracketing is also much less than that in the simplex algorithm. The reason we used the numbers of iterations to measure computation speed is that the abso­lute computation tirne for either algorithm on our computer (Indigo, Silicon Graphics Inc., Mountain View, Calif. 94043) is actually very short since the search starts from close to the minimum cost. This comparison is to show that the better accuracy of successive minimization along coordinate directions as discussed pre­viously is achieved without sacrificing computa­tion speed. on the distance transformation used to generate the edge distance map. The difference in the edge distance map will be carried into the cost function. The "chessboard" distance transfor­mation is one of the simplest approximations of the Euclidean distance transformation be­cause it assigns the same distance from a pixel to ali its 8 immediate neighbors while the Euclidean distance from a pixel to its diagonal neighbor is Closer approximations can be achieved by (a) assigning different numbers to the orthogonal and the diagonal neighbors that better represent the 1 : ff ratio, e.g. the 3 x 3 kernel in the "3-4" distance transformation24 3 O 3 (8) 4 3 4 [4 4] Wang H and Fallone B G or by (b) using a larger kernel to take more neighbors into account, such as the "5-7-11" kernel24 14 11 10 11 14 11 7 5 7 11 10 5 O 5 10 (9) 11 7 5 7 11 14 11 10 11 14 The deviation from the true Euclidean gauge may also result in local minima of the cost function when minimization starts far from the globa! minimum. Except far orientation, the treatment field contour in our case has been brought very close to the matched position by the first approximation [Eq. ( 6)]. To de termine whether a closer approximation to the Eucli­dean distance transfarmation would have an advantage in our case, we compared the effect of the "5-7-11" kernel with that of the "chess­board" kernel. The average edge distance obtai­ned with the "5-7-11" kernel has been divided by 5 to normalize the unit distance to one. Figure 3c shows that the "chessboard" kernel has a slightly better perfarmance than the "5-7­without sacrificing computation speed when matching simple features. In this kind of tasks, successive line minimization of the cost function along coordinate directions is viable when chamfer matching fallows a preliminary align­ment. Appendix: Minimization by bracketing Given a continuous unimodal functionf(x) (Fig. 5) in the region (a, b), the minimum can bereached by simple bracketing with a pair ofpoints a < x\0 < b and a < x\"l < b, wheref (x}I)) < f (x\")). Here the superscripts at l and h denote "low" and "high", respectively, and the subscript i is the iteration number. Starting from an initial pair of points xb') and xbh), the function is evaluated at the reflection point of xb"l about xbl), which is x* = ·xb") + 2 (xb') xb"l), and the value of the function at this point f (x*) is compared with f(xbl)) and f(xb")). Then the fallowing assignments are made: III • (hi .!II f( (II) .r 1 = .r anc J .r 1 = .10 , 1 ·r1·· p· ') < .r0 : ·( (/)) 1·( • ·( ("') and .r ./ .rn ::; .r ) < J .r0 : 11" one. This can be explained rather simply. An edge distance map generated by a distance transfarmation ("chessboard" or "5-7-11") is ,ind .10 . 1 ·-lo . hi) < _ ·1 .1 .• - . 1 + :, very accurate in specifying the orientation of a A precision value e is preset at 104 far our feature. When the minimization with respect to calculations, and if !(r'.'')-f(,:'") the rotation angle is completed, the treatment /( r;") field contour is very close to the fina! registra­ tion. The advantage of the "5-7-11" kernel the process stops; otherwise the process conti­only occurs when the searching process starts nues with another iteration. The programing at a point which is far from registration. This syntax is shown below: situation does not occur in our case because we 1 lh iteration start with a relatively close approximation. On III .r1 . and termmatc; . 1 the other band, the slopes in the edge map generated with the "5-7-11" kernel are steeper so that shaping eITors and noise on the treat­r' = .r;h) + :z(.r;l) .r;h)); ment field contour will have more weight in the · · · , · ( 111) !II lh 1 it J(.r ) < f .r, . t J 1en .r1 +1 = . .i-1 _1 . cost function weakening the correspondence of .• ·( (/)) / 1· • f( Jhl) J 1/J (1) ,,., 1 :.r1 the minimum of the cost function to the actual csc it .f .r, ::::. \.r) < . .r1 , t1cn .r1 +1 = • .r, ..... 1 ('1)) match. In conclusion, our results show that incorpo­rating additional geometric infarmation into (/J ({) 1 (' ti) "' '1 (h) . anJ tcmrnwtc: (h) (<, l-1 ( ,:';', l l 1 ' . . 2: <, thcn .r,,1111 · .r chamfer technique may improve performance clsc. 1 + :!th itcration. A simplified technique for aligning radiation fields in portal imaging f(x) • L 1 1.x a x* x C:: 3:: o . (') t"' u:i (') tTl til tTl v z ('j :1 tTl . Cl o 9 >Tj u:i til >--l > l:ll . . >--l :,:, 5 5 :,:: z >--l f) "rj tTl :'? r') [/) >' tTl t,tTl 0 <.,> N .o - '-O \O EUROPEAN CONGRESS '-O '-O \O --..} O\ '-O OF RADIOLOGY MARCH 2-7, 1997 VIENNA, AUSTRIA :z: -Jv ECR'97 Nepotrebno je, da bolezen spremlja bolecina Moc opioidnega analgetika brez opioidnih stranskih ucinkov centl"alno delujoci analgetik za lajšanje zmet"nih in hudih bolecin ucinkovit ob sot"azmel"no malo stt"anskih ucinkih mocne do mocne akutne ali kronicne bolecine. Po tristopenjski shemi Suetot:ue zdmcs/l'<'U<: 01.Qa11i:::acije .:;:o h(iša11je bolecin pri bolnllolcnfrtw{ia srednje hudo lmlec'lno ali bo{ec'inu dn18<' slo/>1(/<'. Kontraindikacije: Zdravila ne smemo dajati otrokom. mlajšim od 1 leta. Tramadola ne smemo uporabljati pri akutni zastrupitvi z alkoholom, uspavali, analgetiki in drugimi zdravili, ki delujejo na osrednje živcevje. ivled nosecnostjo predpišemo tramaclol lc pri nujni indikaciji. Pri zdravljenju med dojenjem moramo upoštevati, da O, 1 !Jii zdravila prehaja v materino mleko. Pri bolnikih z zvecano obcutljivostjo za oplate moramo tramadol uporabljati zelo previdno. Bolnike s krci centralnega izvora moramo med zdravljenjem skrbno n"dzorovati. Interakcije: Tramadola ne smemo uporabljati skupaj z inhibitorji MAO. Pri soca;-;ni uporabi zdravil, ki delujejo na osrednje živcevje, je možno sinergisticno delovanje v obliki povecane sedacije, pa tudi ugodnejšega analgcticnega delovanja. Opozorila: Pri predoziranju lclhko pride do depresije dihanj;:L Previdnost je potrebna pri bolnikih, ki so preobcutljivi za opiate, pri starejših osebah, pri miksedemu in hipotiroidizmu. Pri okvari jeter in ledvic je potrebno odmerek zmanjšati. Bolniki med zdravljenjem upravljati strojev in motornih vozil. Doziranje in nacin uporabe: Odrasli in otroci, stctrefii 50 do 100 mg i.v.,i.m.,s.c.; intravensko injiciramo pocasi ali infundiramo razredceno v infuzijski Kapsule: 1 kapsula z malo tekoc'ine. Kapljice: 20 kapljic z malo tekocine ali na kocki sladkorja; zadovoljivega u<':inka, dozo ponovimo <':ez 30 do 60 minut. Svecke: 1 .svecka; ce ni ucinka, dozo ponovimo po 3 do 5 urah. Otroci od 1 do 11 let: 1 do 2 mg na kg tele.sne mase. Dnevna doza pri vseh oblikah ne bi smela biti višja od 400 mg. Stranski ucinki: Znojenje, vrtoglavica, slabost, bruhanje, suha usta in utrujenost. Redko lahko pride do palpitacij, ortostatske hipotenzije ali kardiovaskularnega kolapsa. Izjemoma se lahko pojavijo konvulzije. Oprema: 5 ampul po l ml ( 'iO mg/1111), 5 ampul po 2 ml (100 mg/2 ml), 1 O ml raztopine ( 100 mg/ml), 20 kapsul po 50 mg, 5 sveck po JOO mg. Podrobnej.k i1{lormacue so Jl(l uo(jo j)ri J>roiZl!C{jttlcu. ... KRK. SLOVENIJA lopami,[Q 150 -200 -300 -370 mgl/ml. FOR ALL RADIOLOGICAL EXAMINATIONS MYELOGRAPHY ANGIOGRAPHY UROGRAPHY C.T. D.S.A. THE FlaST WAT.ER SOLUBLE READV TO USE NON.IONIC CONiTRAST MEDIUM Manufacturer: Distributer: Bracco s.p.a. Agorest s.r.l. Via E. Folli, 50 Via S. Michele, 334 20134 -Milan -(1) Fax: (02) 26410678 Telex: 311185 Bracco 1 Phone: (02) 21771 e 34170 -Gorizia -(1) Fax: (0481) 20719 Telex: 460690 AF-GO 1 Phone: (0481) 21711 1 Representative Olfice London 5 7-39 'Eastcheap London EC3M 1 DT, U. K. Tel: (+44171) 929 2174 Fax: ( +44 171) 929 2175 OQ9 General Sponsor or 1he Slovcnian Ol_yrnpic Team SKB Banka d.d.: Slovenia's Bank for Glo bal Business SRB Banka d. d. , Ljubljana, Slo­ venia, is a financial services group engaged in banking and imrest­ ment business. It is the largest pri­ vate bank in Sloven:ia. It combines the functions of retail and corporate banking activities through more t11an 40 branches in Slovenia, with a complete range of trade finance and international bank:ing operations, property fi­ nancing, mortgage loans and leas­ ing business. In addition, the bank offers'invest­ ment banking pro'duc.s-and ser­ vices, including trade in secW'ities and custodial services. SKB Banka d. d. symbolizes: -commitment to growth and pros­ perily -a strong capital base and balan­ ce sheet -a clear vision of future develop­ ment -innovative and competitive pro­ ducts and services -flexibility Group SKB: Servi ces, offered by the Bank, are complemented by the following subsidiary companies / affiliates: SKB Real Estate & Leasing Ltd., CA SKB Leasing, SKB Investment Company Ltd., SKB Aurum Ltd., RI-NA Ltd., PLASIS Ltd., Atena lnvestment Fund Managing Co. Ltd., SKB Trading Ltd. jl!J.· uu.-..,1iv, .1u11'1111u SKB BANKA D,D, SI' .-/;'. ' _ ·---------) ________\ _ ) __ """'­ Figure 1. Double-barrelled type carina! reconstruction with both stem bronehi (without pulmonary resection). more distally, the other one in the left stem bronchus was removed, and anastomosis com­pleted. Discussion As it has already been asserted, most lung cancers invading the carina are diagnosed at a stage at which curative resection is no longer possible and other palliative procedures such as laser recanalisation or tracheobronchial intuba­tion are known to have their own limitations. 1 Although the evolution of tracheobronchial re­constructive surgery over the past 25 years has f (--v-l J.. Figure 2. Double-barrelled type carina! reconstruction with right lower and left stem bronchus (in combina­tion with right upper bilobectomy). Carina! resection and reconstruction double-barrelled type References l. Hetzel MR, Smith SGT. Endoscopic palliation of tracheobronchial malignances. Thorax 1991; 46: 325-33. 2. Grillo HC. Carina! reconstruction. Ann Thorac Surg 1982; 34: 356-73. 3. Mathisen DJ, Grillo HC. Carina! resection for bronchogenic carcinoma. J Thorac Cardiovasc Surg 1991; 102: 16-23. 4. Perelman MI, Koroleva NS. Primary tumors of the trachea. In: Grillo HC, Eschapasse H eds. Int Trends Gen Thorac Surg. Philadelphia: Saunders, 1987; 2: 91-106. 5. Mathisen DJ, Grillo HC. Laringotracheal trauma. Acute and chronic injuries. In: Grillo HC, Escha­passe H eds. Int Trends Gen Thorac Surg. Philadel­phia: Saunders, 1987; 2: 117-23. 6. Montgomery WW. Suprahyoid release for tracheal anastomosis. Arch Otolaryngol 1974; 99: 255-60. 7. Pearson FG, Todd TRJ, Cooper JD. Experience with primary neoplasms of the trachea and carina. J Thorac Cardiovasc Surg 1984; 88: 511-18. 8. Grillo HC, Mathisen DJ. Primary traceal tumors: treatment and results. Ann Thorac Surg 1990; 49: 69-77.