- Email: [email protected]
Imaging of the thorax in the management of germ cell testicular tumours
Clinical
Radiology
(1999)
54, 207-211
Imaging of the Thorax in the Management of Germ Cell Testicular Tumours P. M. WHITE*,
D. J. A. ADAMSON?,
G. C. W. HOWARD?,
A. R. WRIGHTS
Departments of *Radiology and iOncology, Western General Hospital, Edinburgh and $Depautment of Radiology, St Mary’s Hospital, London Received: 11 September 1998 Revised: 9 November 1998 Accepted: 28 November 1998 AIM: To evaluate role of chest computed tomography (CTC) and chest radiography (CXR) in management of patients with testicular germ ceil tumours (GCT). PATIENTS AND METHODS: An analysis was undertaken of staging and re-assessment CTC and CXR examinations performed on patients with GCT over a 4.5year period. Data were obtained on clinical presentation, tumour histology, tumour marker levels and clinical course. Consensus review interpretation was combined with these data to obtain a ‘standard of reference’. Sensitivity, specificity and predictive values were derived by comparison of original imaging reports to ‘standard of reference’. RESULTS: Six hundred and twenty-three CTC examinations on 207 patients with GCT were included. Intrathoracic metastases were identified in 1% of seminoma patients compared with 20 % of non-seminoma (NSGCT) patients. CTC was more accurate than CXR in the detection of intrathoracic metastases at 0.97,0.96-0.98 (95% CI) compared with 0.91,0.89-0.93. The agreement between imaging techniques and the standard of reference (determined by Kappa statistic) was respectively 0.96 for CTC and 0.65 for CXR. In GCT patients undergoing re-assessment with both CXR and CTC, CXR never detected unknown intrathoracic metastatic disease. Abdominopelvic lymphadenopathy was associated with intrathoracic metastases (P < O.OOl), however re-assessment CTC did identify intrathoracic metastases in 27 cases without concurrent abdominopelvic disease. CXR was negative in 19 of these. CONCLUSION: Routine interval CXRs are unnecessary in NSGCT patients undergoing regular re-assessment CTC due to the low additional yield and limited effect on management. Re-assessment should still include CTC. In low risk, pure seminoma patients (abdominal CT and marker negative) re-assessment CTC can be safely avoided. Baseline CTC is advocated with CXR alone for re-assessment. White, P. M. et al. (1999) Clinical Radiology 54, 207-211. Key words: computed tomography (CT), utilization, teratoma, testis.
Testicular cancer is an important condition because, although relatively uncommon (< 1% of all male cancers), its incidence is increasing, it mainly affects young men, and carries a good prognosis when treated appropriately. Radiological techniques play a crucial role in the initial staging and subsequent followup of patients with testicular cancer [ 1,2]. Testicular cancer can spread via the lymphatics or via the bloodstream. There is evidence that seminoma is more likely to spread by the lymphatic route; non-seminomatous germ cell tumour (NSGCT) of the testis has a more variable pattern of spread [3]. The detection of chest metastases (either nodal or lung) is an essential part of the staging and re-assessment of patients with testicular tumours [4]. The relative clinical utility of chest radiograph (CXR) and computed tomography of the chest (CTC) and the optimum Correspondence to: Dr P. M. White, X-Ray, Department of Clinical Neurosciences, Western General Hospital, Crewe Road, Edinburgh, EH4 2XU, U.K. 0009-9260/99/040207+05
$12.00/O
re-assessmentschedule for stage I NSGCT and some seminoma patients are controversial with practice varying considerably between centres [5]. It has been the practice at this centre to obtain a CXR as well as CTC at the time of staging and to obtain interval CXRs in patients undergoing intensive re-assessment using regular CT examinations. Typically this regimen is 3-monthly CTC for a year, and then 6-monthly for a further year; seven CT examinations in all. It is uncertain from review of the literature whether this frequency of chest CT examination is necessary [6]. Chest radiographs are obtained monthly for a year, then 2-3 times a month for 2 years and 6 monthly thereafter whilst under regular clinical review. This study aims to evaluate the role of these examinations in the management of patients with germ cell testicular tumours (GCTT). It also incorporates a quantitative assessment of the radiation burden imposed by this imaging activity. 0 1999 The Royal
College
of Radiologists
208 Table
CLINICAL
1 - Analysis
of intrathoracic
disease
Parenchymal Seminoma NSGCTI
by site and underlying
lung mets
1 55
PATIENTS
AND
RADIOLOGY
pathology
Intrathoracic 0 10
METHODS
A retrospective analysis was undertaken of 677 staging, surveillance and follow-up CTC examinations performed on 236 patients with testicular cancer over a 4.5year period at a regional oncology centre. Patients with testicular tumours of non-germ cell histological type and germ cell tumours not of testicular origin were excluded (44 CTC examinations on 21 patients). Patients in whom the case notes were unobtainable were also excluded (10 CTC examinations on eight patients). Therefore, 623 examinations on 207 patients were included in the analysis. Stage I and IIa seminoma patients normally receive only a staging CT examination. However, some of this group did undergo additional follow-up examinations. Typical reasons for this were failure to complete the full course of adjuvant dog leg radiotherapy, and relapse requiring systemic chemotherapy. All CT examinations were performed on a Siemens Somatom Plus CT system (Siemens, Erlangen, Germany). The standard imaging protocol used was spiral CT of the chest from the lung apices to the bases using kV of 137, anmA of 145, 1Omm collimation, a pitch of 1.0-1.5 (the later 180 studies utilized the lower dose protocol of an increased pitch) and reconstruction at 1Omm intervals. Intravenous contrast enhancement was not used routinely. The CXR performed at the time of staging and any interval radiographs performed within a month of a CTC examination during subsequent follow-up were reviewed. CXR were performed using a high kV and 12ft air gap technique. From the clinical notes, data were obtained on clinical presentation, tumour histology, tumour marker levels and the clinical course. Assessment of nodal involvement was made using standard size criteria for short-axis diameter [7,8]. Lung parenchymal metastases were regarded as present if spherical, soft-tissue density lesions which were not demonstrably vesselswere seen in the pulmonary parenchyma. The demonstration of three or fewer pulmonary nodules of < 1 cm diameter on staging examination was not considered diagnostic of pulmonary metastatic disease [9]. In such cases,the presence of nodules was reported as ‘possible metastases’ and close follow-up by CT advised. The presence of more than three nodules of any size was regarded as suspicious for metastases. The presence of any other active, non-neoplastic lung disease was also recorded. Isolated pleural thickening was disregarded. Chest radiographs were analysed using standard criteria. No correlative histological information was available in the great majority of cases.Therefore, in order to try and reach an accurate and definitive decision about each reviewed examination, the method of ‘standard of reference’ was employed. This method has been described in detail in the radiological
(numbers
nodal
mets
refer
to CTC
examinations)
Lung + nodal meti 0 10
._i Mets + other lesions 0 2
Other lesions 2 5
literature as a means of overcoming the problem of ascertaining the nature of a lesion seen on any imaging modality where histological correlation is not available [IO-121. Analysis of the CTC and CXR examinations was performed by two radiologists (A.R.W., P.M.W.) blinded to previous reports. This interpretation was then related to the pathological and biochemical data, and information from clinical follow-up to arrive at a standard of reference. Where there was any disagreement between these data, the CTC or CXR was reviewed again and the situation resolved by consensus. To arrive at sensitivity and specificity measurements, the original radiological reports were compared with the standard of reference assessmentas a ‘gold standard’. For the purposes of the study, each CT examination was regarded as a separate event. Dose estimations were obtained for the above examination protocol using an anthropomorphic body phantom containing thermoluminescent dosimeters [13] (Rando phantom - The Phantom Laboratory, New York, U.S.A.) in conjunction with the Department of Medical Physics, University of Edinburgh. The effective dose equivalent (EDE) for a single posteroanterior CXR using the department’s standard high kV technique was also determined. Statistical analysis of the results was performed using a ‘Systat for Windows’ statistics programme (SYSTAT Inc., Evanston, Illinois, U.S.A.). RESULTS
Of the 207 patients, the underlying pathology was NSGCT in 115 (who had a total of 416 CTC examinations; 50 staging and 366 re-assessment) and seminoma in 92 (207 CTC examinations; 69 staging and 138 re-assessment).Based on the standard of reference technique, metastatic intrathoracic disease was present in 77 (12.36%) CTC examinations from 24 patients (11.59%). Of these, only a single CTC examination in one patient with seminoma was identified as having intrathoracic metastases (0.48 and 1.09% respectively) compared with 76 CTC examinations demonstrating intrathoracic metastases in 23 patients with NSGCT (18.27 and 20.0% respectively) - see Table 1. The difference in prevalence of intrathoracic metastases between seminoma and NSGCT reaches statistical significance P < 0.001 (Fisher’s exact test). The performance of the original CTC and CXR examination reports compared to the standard of reference is shown in Table 2. CTC demonstrated 81/85 true-positives giving a sensitivity of 0.95, 0.91-0.99 (95% CI). ‘True-positive’ means demonstration of active intrathoracic disease, either malignant or benign. The specificity of CTC compared with the standard of reference was 0.97, 0.96-0.98, with a positive predictive value (PPV) of 0.84, 0.83-0.85, negative predictive
THORACIC Table
2 - Identification
of active
intrathoracic
pathology
IMAGING (numbers
IN TESTICULAR refer
True-positive Standard review CTC CXR
of reference
original
report
to imaging
209
CANCER examinations)
True-negative
False-positive
False-negative
Metastatic disease Other lesions
71 8
526
12
0
Metastatic disease Other lesions
14 I 21 2
522
1.5
5
528
3
52
Metastatic disease Other lesions
value (NPV) of 0.99, 0.98-l and an accuracy of 0.97, 0.96-0.98. In three of the cases with intrathoracic pathology, no concurrent CXR was available for comparison. Therefore, out of a possible 82 cases with active intrathoracic pathology, CXR detected 29 true-positives, giving a sensitivity of 0.35, 0.25-0.45, a specificity of 0.99, 0.98-1, PPV of 0.91, 0.89-0.93, NPV of 0.91, 0.89-0.93 and an accuracy of 0.91, 0.89-0.93. The CXR report was in agreement with the CTC report in 541/612 cases (88.40%) and with the standard of reference review in 557/612 (91.01%). Kappa value for the level of agreement between CXR and standard of reference was 0.65, 0.62-0.68, which is rated as good agreement [14]. In comparison the initial CTC report agreed with the standard of reference result in 615/623 (98.72%) cases. Kappa value for level of agreement between CTC and standard of reference was better than for CXR at 0.96, 0.95-0.97 indicating very good agreement [ 141. Non-neoplastic intrathoracic lesions were identified on eight CTC examinations: two examinations demonstrating a rib fracture with haematoma; two demonstrating bullae; three demonstrating pulmonary consolidation; and one revealing an arteriovenous malformation. False-negative CTC examinations: five in total. Three cases were missed lung or mediastinal nodal metastases (in two patients). In one patient two errors of interpretation may have delayed salvage chemotherapy. In the other case the lung metastases not reported were felt on review to be inactive, following radical chemotherapy and with persistently negative markers. The last two false-negatives were failure to report lung nodules on staging CTC examinations (one of the two cases had persistently elevated markers), which although they turned out to be benign granulomata could have been metastases, and should have been commented on as necessitating close follow-up.
patient had had radical chemotherapy and the markers were persistently negative. These errors of interpretation were identified on review and no patient received unnecessary therapy. The relationship between abdominopelvic lymphadenopathy and intrathoracic metastases was assessed.No concurrent CT examination of the abdomen was available for five of the CTC examinations and of these five cases, three had intrathoracic metastases. These were therefore excluded from the analysis. Eighteen of the remaining patients with intrathoracic metastases also had enlarged abdominal and/or pelvic lymph nodes, present on 47 CTC examinations (63.51%) c.f. Tables 3a-c. All of the patients with intrathoracic metastases had had abdominal lymph node enlargement demonstrated on CT at some time during the course of their treatment. The presence of abdominopelvic lymphadenopathy was significantly associated with intrathoracic metastases compared to the presence of intrathoracic metastases in patients with a negative abdominal CT - P
DISCUSSION
False positive CTC examinations: 15 in total. All detected lung nodules were reported, which led in 12 patients with one to three lung nodules (< lcm) to a staging examination being reported as ‘possible lung metastases’. As biopsy of such lesions is not routinely performed in the U.K., close surveillance with CT and tumour markers is mandatory [9]. In one case a mediastinal node of just under borderline size on CT size criteria was mistakenly deemed enlarged. Subsequent review identified this error. In one case (two CTC examinations) the superior pericardial recess was misinterpreted as an enlarged node, although this did not alter clinical management as the
The impetus behind this research was the lack of conclusive evidence in the literature on the use of chest imaging modalities in the staging, surveillance and follow-up of patients with germ cell testicular tumours, particularly seminoma. This problem became apparent to two of the authors (G.C.W.H., A.R.W.) while involved in the preparation of a Scottish Intercollegiate Guidelines Network (SIGN) guideline document on the management of patients with testicular cancer [6]. In this cohort, a number of patients (almost 50%) underwent staging CT examination at another hospital before attending the
210
CLINICAL
Table 3a - Relationship between abdominopelvic metastases and the presence of intrathoracic metastases - numbers in tables 3a-c relate to the consensus review of CT examinations (618 in total)
Abdominopelvic adenopathy (at time of CTC examination) NoAlP lymphadenopathy
Intrathoracic metastases
No intrathoracic metastases
41
101
148 (24%)
27 74 (12%)
443 544 (88%)
470 (76%)
Table 3b - Relationship presence of intrathoracic
Abdominopelvic adenopathy (at time of CTC examination) No AlI’ lymphadenopathy
No A&’ lymphadenopathy
metastases patients
and the
Intrathoracic metastases
No intratboracic metastases
46
58
104 (25.18%)
27 73 (17.68%)
282 340 (82.32%)
309 (74.82%)
Table 3c - Relationship presence of intrathoracic
Abdominopelvic adenopathy (at time of CTC examination)
between abdominopelvic metastases for NSGCT
between abdominopelvic metastases for seminoma
metastases patients
and the
Intrathoracic metastases
No intrathoracic metastases
1
43
44 (21.46%)
0 1 (0.49%)
161 204 (99.51%)
161 (78.54%)
Centre for treatment and subsequent re-assessment.In such a case it is good practice for an experienced oncological radiologist to review the staging examination [6]. For consistency of examination technique and logistical reasons, examinations from other hospitals were not included in our review analysis. The cohort may also be somewhat unusual in the number of re-assessment examinations performed in seminoma patients (for reasons outlined earlier). The results confirm previous findings of a low rBte of pulmonary parenchymal metastases in seminoma in the range 0.3 to 3.5% [9,16], 1.1% in this study. Unexpectedly we did not identify any cases of intrathoracic nodal metastases, which have been reported as occurring more frequently than parenchymal lung metastases in seminoma [3,9]. The intrathoracic metastatic rate for NSGCT at 20% is in keeping with previous reports which give a range of 17.0-25.7% [17-191. Importantly, no patient undergoing regular CT surveillance in this
Regional
RADIOLOGY
series had a CXR which detected unknown intrathoracic metastatic disease. Almost one-third of patients with positive abdominal CT examinations had concurrent intrathoracic metastases, again similar to previous reports [ 18,191, with the association reaching statistical significance. For NSGCT, 44.23% of positive CTA examinations were associated with intrathoracic metastases, compared with only 8.74% of negative CTA examinations. The corresponding rates for seminoma are 2.27% and 0% respectively. It has been suggested that CTC provides no useful additional information in patients with a negative CTA [ 181.We identified 27 CTC examinations from 10 patients who had intrathoracic metastases in the absence of concurrent abdominopelvic lymphadenopathy (all with NSGCT). All 10 patients had been marker positive at initial presentation and all had had abdominal lymphadenopathy at some stage. The concurrent CXR was negative in 19 of the 27 cases and markers were elevated in only two cases at time the chest metastases were found. Clinically there was no evidence of disease.Therefore, without a CTC examination, intrathoracic metastatic disease would have been missed on 17 occasions in seven patients. This would have led to a significant delay in treatment for disease relapse in four of these patients. There does appear to be a significant difference in the behaviour of seminoma and NSGCT in the pattern of spread of metastases to the chest, with this being both more frequent and more variable in NSGCT. In the UK, CTC at staging and follow-up is regarded as standard practice without a differentiation in imaging protocol between seminoma and NSGCT [.5,20]. However, CXR alone has been advocated in the North American literature for both the staging and follow-up of seminoma and even NSGCT where the CTA is negative [l&19,21]. Such a policy would not seem to be appropriate for NSGCT with more frequent intrathoracic metastases and a variable pattern of spread. Where the pre-test probability of intrathoracic metastases is relatively high, CTC has been advocated as mandatory [22-241. Our data may support the use of CXR alone in seminoma patients where the pre-test probability of intrathoracic metastases is very low, i.e. CTA and marker negative. However, CT is a more accurate staging tool than CXR in the chest for both detection of disease and defining its extent [9], both of which may be crucial in guiding management and on subsequent follow-up. Therefore, there is a strong casefor performing a baseline CTC at the time of staging even if the pre-test probability of intrathoracic metastatic disease is low. The main disadvantages of CTC are its poor specificity and radiation burden compared to CXR. Sensitivity, particularly for small lung nodules, is better for CTC than CXR. Modern spiral CT machines are readily able to detect 2-3 mm diameter nodules [25], whereas detection of nodules < 5 mm is difficult on CXR (particularly if nodules are few in number). For the purposes of this study any lung nodule detected was regarded as suspicious of pulmonary metastatic disease. Twelve out of fifteen false-positives on CTC in this study were in the category of having three or fewer pulmonary nodules of c 1 cm diameter, and could therefore be predicted to be non-metastatic on accepted criteria [9]. If this well-defined predictable subgroup of ‘false positives’ are removed from the analysis, the
THORACIC
IMAGING
specificity of CTC improves to 0.99, 0.98-1.0 (the same as for CXR), whilst maintaining excellent sensitivity. Increased use of spiral CT at a pitch of up to 1.5 has become accepted practice and reduces the EDE by one-third without affecting diagnostic accuracy [25]. If follow-up CTC were not performed in seminoma patients with a low pre-test probability of intrathoracic metastases, up to 21% (103/499 re-assessment CTC examinations from this series)would not be required. This would have reduced the cohort EDE in this study by 0.45 man Sv. In practice as the lung bases would be included in an examination of the abdomen (amount of overlap is usually in the range 4-8 cm), the EDE would not be reduced by as much as this. Nevertheless, a useful dose reduction for some patients can be safely achieved in line with the ALARA principle. Reduction of unnecessary CTC and CXR examinations also offers a significant resource saving. Acknowledgements. The authors would like to thank Mr .I. R. Williams for his help with obtaining the dosimetry data and for his helpful comments on the study.
REFERENCES Horwich A. Testicular germ cell tumours:an introductory overview. In: Horwich A, ed. Testicular Cancer. London: Chapman and Hall, 1991;1-13. Roth BJ, Nichols CR. Testicular cancer in the 90’s: Victory and a new call to arms. Sem Orzcol 1992;19:117-118. Wood A, Robson N, Ting K et al. Patterns of supradiaphragmatic metastases in testicular germ cell tumours. Clin Radio1 1996;51:273-276. Mead G. Testis. In: Price P, Sikora K, eds. Treatment of Cancer. London: Chapman and Hall, 1995;627-645. MacVicar D. Staging of testicular germ cell tumours. Clin Radio1 1993;47:149-158. Scottish Intercollegiate Guidelines Network. Management of Adulf Testicular Germ Cell Turnours. 1999 (in press). Dorfman RE, Alpern MB, Gross BH, et al. Upper abdominal lymph nodes: criteria for normal size determined with CT. Radiology 1991;180:319-322. Glazer GM, Gross BH, Quint LE, et al. Normal mediastinal lymph nodes: number and size according to American Thoracic Society mapping. Am .l Roentgen01 1985;144:261-265.
IN TESTICULAR
CANCER
211
9 Williams M, Husband .I, Heron CW. Intrathoracic manifestations of metastatic testicular seminoma: a comparison of chest radiographic and CT findings. Am J Roentgen01 1987;149:473-475. 10 White PM, Howard GCW, Best JJK, et al. The role of Computed Tomographic examination of the pelvis in the management of testicular germ cell tumours. C2in Radial 1997;52:124-129. 11 Haubold-Reuter BG, Duewell S, Schilcher BR, et al. The value of bone scintigraphy, bone marrow scintigraphy and fast spin-echo magnetic resonance imaging in the staging of patients with malignant solid turnouts: a prospective study. Eur J Nucl Med 1993;20:10631069. 12 Kosuda S, Kaji J, Yokoyama H, et al. Does bone SPECT actually have lower sensitivity for detecting vertebral metastasis than MRI? / Nucl Med 1996;37:975-978. 13 Collie DA, Wright AR, Williams JR, et al. Comparison of spiral acquisition computed tomography and conventional computed tomography in the assessment of pulmonary metastatic disease. Br J Radiology 1994;67:436-444. Stutistics 14 Altman DG. Inter rater agreement. In: Altman DG Practical for Medical Research. London: Chapman and Hall, 1991; Chapter 14. 15 International Commission of Radiological Protection. Recommendutions of the ZCRP. Oxford: Pergamon Press, 1991;1-201. 16 Zwaveling A, Soebhag R. Testicular tumours in the Netherlands. Cancer 1985;55:1612-1617. 17 Lien H, Lindskold L, Fossa S, et al. Computed tomography and conventional radiography in intrathoracic metastases from non seminomatous testicular tumour. Acta Rudiologica (Diagnosis) 1988;29:547-549. 18 Moul J. Chest and Abdominal Staging in Testicular Cancer Patients. Med Imug ht 1995;5:14-18. 19 See W, Hoxie L. Chest staging in testis cancer patients: imaging modality selection based upon risk assessment as determined by abdominal computerized tomography scan results. J Urolo 1993;150:874-878. 20 Council RCR. The Use of Computed Tomography in the Initial Investigation of Common Malignancies. London: Royal College of Radiologists, 1994;25-25. 21 Steinfeld A. Macher M. Radiologic staging of chest in testicular seminoma. Urology 1990;36:428-GO. - of pulmonary nodules by 22 Muhm J, Brown L, Crowe I. Detection computed tomography. Am J Roentgen01 1977;128:267-270. of pulmonary metastasis in patients with 23 Davis S. CT evaluation extrathoracic malignancy. Radiology 1991;180: l-l 2. 24 Weinstein M, Finberg H. Clinical Decision Analysis. Philadelphia: WB Saunders Co., 1980. JR, et al. Pulmonary nodules: effect 25 Wright AR, Collie DA, Williams on detection of spiral CT pitch. Radiology 1996;199:837-841.
Radiology
(1999)
54, 207-211
Imaging of the Thorax in the Management of Germ Cell Testicular Tumours P. M. WHITE*,
D. J. A. ADAMSON?,
G. C. W. HOWARD?,
A. R. WRIGHTS
Departments of *Radiology and iOncology, Western General Hospital, Edinburgh and $Depautment of Radiology, St Mary’s Hospital, London Received: 11 September 1998 Revised: 9 November 1998 Accepted: 28 November 1998 AIM: To evaluate role of chest computed tomography (CTC) and chest radiography (CXR) in management of patients with testicular germ ceil tumours (GCT). PATIENTS AND METHODS: An analysis was undertaken of staging and re-assessment CTC and CXR examinations performed on patients with GCT over a 4.5year period. Data were obtained on clinical presentation, tumour histology, tumour marker levels and clinical course. Consensus review interpretation was combined with these data to obtain a ‘standard of reference’. Sensitivity, specificity and predictive values were derived by comparison of original imaging reports to ‘standard of reference’. RESULTS: Six hundred and twenty-three CTC examinations on 207 patients with GCT were included. Intrathoracic metastases were identified in 1% of seminoma patients compared with 20 % of non-seminoma (NSGCT) patients. CTC was more accurate than CXR in the detection of intrathoracic metastases at 0.97,0.96-0.98 (95% CI) compared with 0.91,0.89-0.93. The agreement between imaging techniques and the standard of reference (determined by Kappa statistic) was respectively 0.96 for CTC and 0.65 for CXR. In GCT patients undergoing re-assessment with both CXR and CTC, CXR never detected unknown intrathoracic metastatic disease. Abdominopelvic lymphadenopathy was associated with intrathoracic metastases (P < O.OOl), however re-assessment CTC did identify intrathoracic metastases in 27 cases without concurrent abdominopelvic disease. CXR was negative in 19 of these. CONCLUSION: Routine interval CXRs are unnecessary in NSGCT patients undergoing regular re-assessment CTC due to the low additional yield and limited effect on management. Re-assessment should still include CTC. In low risk, pure seminoma patients (abdominal CT and marker negative) re-assessment CTC can be safely avoided. Baseline CTC is advocated with CXR alone for re-assessment. White, P. M. et al. (1999) Clinical Radiology 54, 207-211. Key words: computed tomography (CT), utilization, teratoma, testis.
Testicular cancer is an important condition because, although relatively uncommon (< 1% of all male cancers), its incidence is increasing, it mainly affects young men, and carries a good prognosis when treated appropriately. Radiological techniques play a crucial role in the initial staging and subsequent followup of patients with testicular cancer [ 1,2]. Testicular cancer can spread via the lymphatics or via the bloodstream. There is evidence that seminoma is more likely to spread by the lymphatic route; non-seminomatous germ cell tumour (NSGCT) of the testis has a more variable pattern of spread [3]. The detection of chest metastases (either nodal or lung) is an essential part of the staging and re-assessment of patients with testicular tumours [4]. The relative clinical utility of chest radiograph (CXR) and computed tomography of the chest (CTC) and the optimum Correspondence to: Dr P. M. White, X-Ray, Department of Clinical Neurosciences, Western General Hospital, Crewe Road, Edinburgh, EH4 2XU, U.K. 0009-9260/99/040207+05
$12.00/O
re-assessmentschedule for stage I NSGCT and some seminoma patients are controversial with practice varying considerably between centres [5]. It has been the practice at this centre to obtain a CXR as well as CTC at the time of staging and to obtain interval CXRs in patients undergoing intensive re-assessment using regular CT examinations. Typically this regimen is 3-monthly CTC for a year, and then 6-monthly for a further year; seven CT examinations in all. It is uncertain from review of the literature whether this frequency of chest CT examination is necessary [6]. Chest radiographs are obtained monthly for a year, then 2-3 times a month for 2 years and 6 monthly thereafter whilst under regular clinical review. This study aims to evaluate the role of these examinations in the management of patients with germ cell testicular tumours (GCTT). It also incorporates a quantitative assessment of the radiation burden imposed by this imaging activity. 0 1999 The Royal
College
of Radiologists
208 Table
CLINICAL
1 - Analysis
of intrathoracic
disease
Parenchymal Seminoma NSGCTI
by site and underlying
lung mets
1 55
PATIENTS
AND
RADIOLOGY
pathology
Intrathoracic 0 10
METHODS
A retrospective analysis was undertaken of 677 staging, surveillance and follow-up CTC examinations performed on 236 patients with testicular cancer over a 4.5year period at a regional oncology centre. Patients with testicular tumours of non-germ cell histological type and germ cell tumours not of testicular origin were excluded (44 CTC examinations on 21 patients). Patients in whom the case notes were unobtainable were also excluded (10 CTC examinations on eight patients). Therefore, 623 examinations on 207 patients were included in the analysis. Stage I and IIa seminoma patients normally receive only a staging CT examination. However, some of this group did undergo additional follow-up examinations. Typical reasons for this were failure to complete the full course of adjuvant dog leg radiotherapy, and relapse requiring systemic chemotherapy. All CT examinations were performed on a Siemens Somatom Plus CT system (Siemens, Erlangen, Germany). The standard imaging protocol used was spiral CT of the chest from the lung apices to the bases using kV of 137, anmA of 145, 1Omm collimation, a pitch of 1.0-1.5 (the later 180 studies utilized the lower dose protocol of an increased pitch) and reconstruction at 1Omm intervals. Intravenous contrast enhancement was not used routinely. The CXR performed at the time of staging and any interval radiographs performed within a month of a CTC examination during subsequent follow-up were reviewed. CXR were performed using a high kV and 12ft air gap technique. From the clinical notes, data were obtained on clinical presentation, tumour histology, tumour marker levels and the clinical course. Assessment of nodal involvement was made using standard size criteria for short-axis diameter [7,8]. Lung parenchymal metastases were regarded as present if spherical, soft-tissue density lesions which were not demonstrably vesselswere seen in the pulmonary parenchyma. The demonstration of three or fewer pulmonary nodules of < 1 cm diameter on staging examination was not considered diagnostic of pulmonary metastatic disease [9]. In such cases,the presence of nodules was reported as ‘possible metastases’ and close follow-up by CT advised. The presence of more than three nodules of any size was regarded as suspicious for metastases. The presence of any other active, non-neoplastic lung disease was also recorded. Isolated pleural thickening was disregarded. Chest radiographs were analysed using standard criteria. No correlative histological information was available in the great majority of cases.Therefore, in order to try and reach an accurate and definitive decision about each reviewed examination, the method of ‘standard of reference’ was employed. This method has been described in detail in the radiological
(numbers
nodal
mets
refer
to CTC
examinations)
Lung + nodal meti 0 10
._i Mets + other lesions 0 2
Other lesions 2 5
literature as a means of overcoming the problem of ascertaining the nature of a lesion seen on any imaging modality where histological correlation is not available [IO-121. Analysis of the CTC and CXR examinations was performed by two radiologists (A.R.W., P.M.W.) blinded to previous reports. This interpretation was then related to the pathological and biochemical data, and information from clinical follow-up to arrive at a standard of reference. Where there was any disagreement between these data, the CTC or CXR was reviewed again and the situation resolved by consensus. To arrive at sensitivity and specificity measurements, the original radiological reports were compared with the standard of reference assessmentas a ‘gold standard’. For the purposes of the study, each CT examination was regarded as a separate event. Dose estimations were obtained for the above examination protocol using an anthropomorphic body phantom containing thermoluminescent dosimeters [13] (Rando phantom - The Phantom Laboratory, New York, U.S.A.) in conjunction with the Department of Medical Physics, University of Edinburgh. The effective dose equivalent (EDE) for a single posteroanterior CXR using the department’s standard high kV technique was also determined. Statistical analysis of the results was performed using a ‘Systat for Windows’ statistics programme (SYSTAT Inc., Evanston, Illinois, U.S.A.). RESULTS
Of the 207 patients, the underlying pathology was NSGCT in 115 (who had a total of 416 CTC examinations; 50 staging and 366 re-assessment) and seminoma in 92 (207 CTC examinations; 69 staging and 138 re-assessment).Based on the standard of reference technique, metastatic intrathoracic disease was present in 77 (12.36%) CTC examinations from 24 patients (11.59%). Of these, only a single CTC examination in one patient with seminoma was identified as having intrathoracic metastases (0.48 and 1.09% respectively) compared with 76 CTC examinations demonstrating intrathoracic metastases in 23 patients with NSGCT (18.27 and 20.0% respectively) - see Table 1. The difference in prevalence of intrathoracic metastases between seminoma and NSGCT reaches statistical significance P < 0.001 (Fisher’s exact test). The performance of the original CTC and CXR examination reports compared to the standard of reference is shown in Table 2. CTC demonstrated 81/85 true-positives giving a sensitivity of 0.95, 0.91-0.99 (95% CI). ‘True-positive’ means demonstration of active intrathoracic disease, either malignant or benign. The specificity of CTC compared with the standard of reference was 0.97, 0.96-0.98, with a positive predictive value (PPV) of 0.84, 0.83-0.85, negative predictive
THORACIC Table
2 - Identification
of active
intrathoracic
pathology
IMAGING (numbers
IN TESTICULAR refer
True-positive Standard review CTC CXR
of reference
original
report
to imaging
209
CANCER examinations)
True-negative
False-positive
False-negative
Metastatic disease Other lesions
71 8
526
12
0
Metastatic disease Other lesions
14 I 21 2
522
1.5
5
528
3
52
Metastatic disease Other lesions
value (NPV) of 0.99, 0.98-l and an accuracy of 0.97, 0.96-0.98. In three of the cases with intrathoracic pathology, no concurrent CXR was available for comparison. Therefore, out of a possible 82 cases with active intrathoracic pathology, CXR detected 29 true-positives, giving a sensitivity of 0.35, 0.25-0.45, a specificity of 0.99, 0.98-1, PPV of 0.91, 0.89-0.93, NPV of 0.91, 0.89-0.93 and an accuracy of 0.91, 0.89-0.93. The CXR report was in agreement with the CTC report in 541/612 cases (88.40%) and with the standard of reference review in 557/612 (91.01%). Kappa value for the level of agreement between CXR and standard of reference was 0.65, 0.62-0.68, which is rated as good agreement [14]. In comparison the initial CTC report agreed with the standard of reference result in 615/623 (98.72%) cases. Kappa value for level of agreement between CTC and standard of reference was better than for CXR at 0.96, 0.95-0.97 indicating very good agreement [ 141. Non-neoplastic intrathoracic lesions were identified on eight CTC examinations: two examinations demonstrating a rib fracture with haematoma; two demonstrating bullae; three demonstrating pulmonary consolidation; and one revealing an arteriovenous malformation. False-negative CTC examinations: five in total. Three cases were missed lung or mediastinal nodal metastases (in two patients). In one patient two errors of interpretation may have delayed salvage chemotherapy. In the other case the lung metastases not reported were felt on review to be inactive, following radical chemotherapy and with persistently negative markers. The last two false-negatives were failure to report lung nodules on staging CTC examinations (one of the two cases had persistently elevated markers), which although they turned out to be benign granulomata could have been metastases, and should have been commented on as necessitating close follow-up.
patient had had radical chemotherapy and the markers were persistently negative. These errors of interpretation were identified on review and no patient received unnecessary therapy. The relationship between abdominopelvic lymphadenopathy and intrathoracic metastases was assessed.No concurrent CT examination of the abdomen was available for five of the CTC examinations and of these five cases, three had intrathoracic metastases. These were therefore excluded from the analysis. Eighteen of the remaining patients with intrathoracic metastases also had enlarged abdominal and/or pelvic lymph nodes, present on 47 CTC examinations (63.51%) c.f. Tables 3a-c. All of the patients with intrathoracic metastases had had abdominal lymph node enlargement demonstrated on CT at some time during the course of their treatment. The presence of abdominopelvic lymphadenopathy was significantly associated with intrathoracic metastases compared to the presence of intrathoracic metastases in patients with a negative abdominal CT - P
DISCUSSION
False positive CTC examinations: 15 in total. All detected lung nodules were reported, which led in 12 patients with one to three lung nodules (< lcm) to a staging examination being reported as ‘possible lung metastases’. As biopsy of such lesions is not routinely performed in the U.K., close surveillance with CT and tumour markers is mandatory [9]. In one case a mediastinal node of just under borderline size on CT size criteria was mistakenly deemed enlarged. Subsequent review identified this error. In one case (two CTC examinations) the superior pericardial recess was misinterpreted as an enlarged node, although this did not alter clinical management as the
The impetus behind this research was the lack of conclusive evidence in the literature on the use of chest imaging modalities in the staging, surveillance and follow-up of patients with germ cell testicular tumours, particularly seminoma. This problem became apparent to two of the authors (G.C.W.H., A.R.W.) while involved in the preparation of a Scottish Intercollegiate Guidelines Network (SIGN) guideline document on the management of patients with testicular cancer [6]. In this cohort, a number of patients (almost 50%) underwent staging CT examination at another hospital before attending the
210
CLINICAL
Table 3a - Relationship between abdominopelvic metastases and the presence of intrathoracic metastases - numbers in tables 3a-c relate to the consensus review of CT examinations (618 in total)
Abdominopelvic adenopathy (at time of CTC examination) NoAlP lymphadenopathy
Intrathoracic metastases
No intrathoracic metastases
41
101
148 (24%)
27 74 (12%)
443 544 (88%)
470 (76%)
Table 3b - Relationship presence of intrathoracic
Abdominopelvic adenopathy (at time of CTC examination) No AlI’ lymphadenopathy
No A&’ lymphadenopathy
metastases patients
and the
Intrathoracic metastases
No intratboracic metastases
46
58
104 (25.18%)
27 73 (17.68%)
282 340 (82.32%)
309 (74.82%)
Table 3c - Relationship presence of intrathoracic
Abdominopelvic adenopathy (at time of CTC examination)
between abdominopelvic metastases for NSGCT
between abdominopelvic metastases for seminoma
metastases patients
and the
Intrathoracic metastases
No intrathoracic metastases
1
43
44 (21.46%)
0 1 (0.49%)
161 204 (99.51%)
161 (78.54%)
Centre for treatment and subsequent re-assessment.In such a case it is good practice for an experienced oncological radiologist to review the staging examination [6]. For consistency of examination technique and logistical reasons, examinations from other hospitals were not included in our review analysis. The cohort may also be somewhat unusual in the number of re-assessment examinations performed in seminoma patients (for reasons outlined earlier). The results confirm previous findings of a low rBte of pulmonary parenchymal metastases in seminoma in the range 0.3 to 3.5% [9,16], 1.1% in this study. Unexpectedly we did not identify any cases of intrathoracic nodal metastases, which have been reported as occurring more frequently than parenchymal lung metastases in seminoma [3,9]. The intrathoracic metastatic rate for NSGCT at 20% is in keeping with previous reports which give a range of 17.0-25.7% [17-191. Importantly, no patient undergoing regular CT surveillance in this
Regional
RADIOLOGY
series had a CXR which detected unknown intrathoracic metastatic disease. Almost one-third of patients with positive abdominal CT examinations had concurrent intrathoracic metastases, again similar to previous reports [ 18,191, with the association reaching statistical significance. For NSGCT, 44.23% of positive CTA examinations were associated with intrathoracic metastases, compared with only 8.74% of negative CTA examinations. The corresponding rates for seminoma are 2.27% and 0% respectively. It has been suggested that CTC provides no useful additional information in patients with a negative CTA [ 181.We identified 27 CTC examinations from 10 patients who had intrathoracic metastases in the absence of concurrent abdominopelvic lymphadenopathy (all with NSGCT). All 10 patients had been marker positive at initial presentation and all had had abdominal lymphadenopathy at some stage. The concurrent CXR was negative in 19 of the 27 cases and markers were elevated in only two cases at time the chest metastases were found. Clinically there was no evidence of disease.Therefore, without a CTC examination, intrathoracic metastatic disease would have been missed on 17 occasions in seven patients. This would have led to a significant delay in treatment for disease relapse in four of these patients. There does appear to be a significant difference in the behaviour of seminoma and NSGCT in the pattern of spread of metastases to the chest, with this being both more frequent and more variable in NSGCT. In the UK, CTC at staging and follow-up is regarded as standard practice without a differentiation in imaging protocol between seminoma and NSGCT [.5,20]. However, CXR alone has been advocated in the North American literature for both the staging and follow-up of seminoma and even NSGCT where the CTA is negative [l&19,21]. Such a policy would not seem to be appropriate for NSGCT with more frequent intrathoracic metastases and a variable pattern of spread. Where the pre-test probability of intrathoracic metastases is relatively high, CTC has been advocated as mandatory [22-241. Our data may support the use of CXR alone in seminoma patients where the pre-test probability of intrathoracic metastases is very low, i.e. CTA and marker negative. However, CT is a more accurate staging tool than CXR in the chest for both detection of disease and defining its extent [9], both of which may be crucial in guiding management and on subsequent follow-up. Therefore, there is a strong casefor performing a baseline CTC at the time of staging even if the pre-test probability of intrathoracic metastatic disease is low. The main disadvantages of CTC are its poor specificity and radiation burden compared to CXR. Sensitivity, particularly for small lung nodules, is better for CTC than CXR. Modern spiral CT machines are readily able to detect 2-3 mm diameter nodules [25], whereas detection of nodules < 5 mm is difficult on CXR (particularly if nodules are few in number). For the purposes of this study any lung nodule detected was regarded as suspicious of pulmonary metastatic disease. Twelve out of fifteen false-positives on CTC in this study were in the category of having three or fewer pulmonary nodules of c 1 cm diameter, and could therefore be predicted to be non-metastatic on accepted criteria [9]. If this well-defined predictable subgroup of ‘false positives’ are removed from the analysis, the
THORACIC
IMAGING
specificity of CTC improves to 0.99, 0.98-1.0 (the same as for CXR), whilst maintaining excellent sensitivity. Increased use of spiral CT at a pitch of up to 1.5 has become accepted practice and reduces the EDE by one-third without affecting diagnostic accuracy [25]. If follow-up CTC were not performed in seminoma patients with a low pre-test probability of intrathoracic metastases, up to 21% (103/499 re-assessment CTC examinations from this series)would not be required. This would have reduced the cohort EDE in this study by 0.45 man Sv. In practice as the lung bases would be included in an examination of the abdomen (amount of overlap is usually in the range 4-8 cm), the EDE would not be reduced by as much as this. Nevertheless, a useful dose reduction for some patients can be safely achieved in line with the ALARA principle. Reduction of unnecessary CTC and CXR examinations also offers a significant resource saving. Acknowledgements. The authors would like to thank Mr .I. R. Williams for his help with obtaining the dosimetry data and for his helpful comments on the study.
REFERENCES Horwich A. Testicular germ cell tumours:an introductory overview. In: Horwich A, ed. Testicular Cancer. London: Chapman and Hall, 1991;1-13. Roth BJ, Nichols CR. Testicular cancer in the 90’s: Victory and a new call to arms. Sem Orzcol 1992;19:117-118. Wood A, Robson N, Ting K et al. Patterns of supradiaphragmatic metastases in testicular germ cell tumours. Clin Radio1 1996;51:273-276. Mead G. Testis. In: Price P, Sikora K, eds. Treatment of Cancer. London: Chapman and Hall, 1995;627-645. MacVicar D. Staging of testicular germ cell tumours. Clin Radio1 1993;47:149-158. Scottish Intercollegiate Guidelines Network. Management of Adulf Testicular Germ Cell Turnours. 1999 (in press). Dorfman RE, Alpern MB, Gross BH, et al. Upper abdominal lymph nodes: criteria for normal size determined with CT. Radiology 1991;180:319-322. Glazer GM, Gross BH, Quint LE, et al. Normal mediastinal lymph nodes: number and size according to American Thoracic Society mapping. Am .l Roentgen01 1985;144:261-265.
IN TESTICULAR
CANCER
211
9 Williams M, Husband .I, Heron CW. Intrathoracic manifestations of metastatic testicular seminoma: a comparison of chest radiographic and CT findings. Am J Roentgen01 1987;149:473-475. 10 White PM, Howard GCW, Best JJK, et al. The role of Computed Tomographic examination of the pelvis in the management of testicular germ cell tumours. C2in Radial 1997;52:124-129. 11 Haubold-Reuter BG, Duewell S, Schilcher BR, et al. The value of bone scintigraphy, bone marrow scintigraphy and fast spin-echo magnetic resonance imaging in the staging of patients with malignant solid turnouts: a prospective study. Eur J Nucl Med 1993;20:10631069. 12 Kosuda S, Kaji J, Yokoyama H, et al. Does bone SPECT actually have lower sensitivity for detecting vertebral metastasis than MRI? / Nucl Med 1996;37:975-978. 13 Collie DA, Wright AR, Williams JR, et al. Comparison of spiral acquisition computed tomography and conventional computed tomography in the assessment of pulmonary metastatic disease. Br J Radiology 1994;67:436-444. Stutistics 14 Altman DG. Inter rater agreement. In: Altman DG Practical for Medical Research. London: Chapman and Hall, 1991; Chapter 14. 15 International Commission of Radiological Protection. Recommendutions of the ZCRP. Oxford: Pergamon Press, 1991;1-201. 16 Zwaveling A, Soebhag R. Testicular tumours in the Netherlands. Cancer 1985;55:1612-1617. 17 Lien H, Lindskold L, Fossa S, et al. Computed tomography and conventional radiography in intrathoracic metastases from non seminomatous testicular tumour. Acta Rudiologica (Diagnosis) 1988;29:547-549. 18 Moul J. Chest and Abdominal Staging in Testicular Cancer Patients. Med Imug ht 1995;5:14-18. 19 See W, Hoxie L. Chest staging in testis cancer patients: imaging modality selection based upon risk assessment as determined by abdominal computerized tomography scan results. J Urolo 1993;150:874-878. 20 Council RCR. The Use of Computed Tomography in the Initial Investigation of Common Malignancies. London: Royal College of Radiologists, 1994;25-25. 21 Steinfeld A. Macher M. Radiologic staging of chest in testicular seminoma. Urology 1990;36:428-GO. - of pulmonary nodules by 22 Muhm J, Brown L, Crowe I. Detection computed tomography. Am J Roentgen01 1977;128:267-270. of pulmonary metastasis in patients with 23 Davis S. CT evaluation extrathoracic malignancy. Radiology 1991;180: l-l 2. 24 Weinstein M, Finberg H. Clinical Decision Analysis. Philadelphia: WB Saunders Co., 1980. JR, et al. Pulmonary nodules: effect 25 Wright AR, Collie DA, Williams on detection of spiral CT pitch. Radiology 1996;199:837-841.