Breakdown of chest radiographyresults.
Kaplan-Meier survival curves forasymptomatic patients converted to stage IV by detection of pulmonary recurrenceon routine chest radiography (C-IV; squares; n = 41); and patients with previouslydocumented stage IV disease who developed further pulmonary involvement asdetected by chest radiography (K-IV; circles; n = 34). P = .68by log-rank test.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Tsao H, Feldman M, Fullerton JE, Sober AJ, Rosenthal D, Goggins W. Early Detection of Asymptomatic Pulmonary Melanoma Metastases by RoutineChest Radiographs Is Not Associated With Improved Survival. Arch Dermatol. 2004;140(1):67–70. doi:10.1001/archderm.140.1.67
To determine if earlier detection of pulmonary metastasis by routinechest radiography (CR) is associated with a prolonged survival.
A computer-assisted search of all CR reports on patients with melanomabetween 1990 and 1994 at the Massachusetts General Hospital, Boston. Positiveor suspicious findings for pulmonary metastasis were further pursued throughreview of medical records and tumor registry files.
Setting and Patients
A hospital-based population of patients with melanoma undergoing routineCR at Massachusetts General Hospital.
Overall, of 994 patients, 75 were identified as having pulmonary metastasesby CR (1937 total chest radiographs). In addition, there were 63 patientswith suspicious findings that were later shown to be false positive. Chestradiographs provided the initial evidence of metastases in 41 asymptomaticindividuals. Thirty-four patients had known melanoma metastases to other sitesat the time of the first abnormal chest radiograph. Survival after identificationof pulmonary metastasis did not differ significantly between the 2 groups.
In this study, there was no evidence to support the notion that earlierdetection of pulmonary metastasis in otherwise asymptomatic individuals confersa survival advantage in an unselected melanoma population.
ALTHOUGH CHEST RADIOGRAphy (CR) is routinely used in the initial stagingof patients with cutaneous melanoma and in postmelanoma surveillance, itsutility and impact on survival is still uncertain. Terhune et al1 reportedthat only 1 of 876 patients had a true-positive CR finding on initial stagingfor localized melanoma, although an additional 128 disease-free patients hadabnormalities that necessitated further workup. This relatively high false-positiverate (15%) adds both a real economic cost to the health care system and apotential emotional cost to the patient.1 Theutility of CR as a component of postmelanoma surveillance has also been challenged,since several studies have found that most melanoma recurrences after localresection are detected through symptoms and/or physical findings rather thanlaboratory tests, such as CR.2-5 Earlierdetection of resectable pulmonary disease may lead to earlier surgery andbetter outcome in some selected patients.2,6-8 However,these individuals represent a highly select group of patients with stage IVdisease referred to surgical centers, and the impact of earlier detectionby routine CR on survival of the general postmelanoma population is stillunknown. The routine use of CR has not been promulgated worldwide. For instance,guidelines from the United Kingdom,9 the Netherlands,10 and Australia11 donot consider radiological examination worthwhile in routine follow-up. Althoughthe German Society of Dermatology recommends annual CR for patients with localizeddisease and biannual CR for patients with in-transit and/or regional disease,12 Garbe et al12 recentlyfound that only 12 of 2396 routine chest radiographs (0.5%) obtained overa 25-month period were confirmed as true metastases.
In the United States, although the National Comprehensive Cancer Network(http://www.nccn.org) and the American Academy of Dermatology13 endorse routine CR only as an option during follow-upsurveillance, a recent survey of 30 dermatologists, surgical oncologists,and medical oncologists revealed that more than 70% of physicians treatingpatients with melanoma ordered annual CR during the first 5 years after initialdiagnosis.14 Thus, routine CR is still partof common practice, although there is no evidence that earlier detection ofasymptomatic pulmonary metastases confers a survival advantage in the generalmelanoma population. To this end, we set out to determine if there is apparentsurvival prolongation associated with earlier detection of asymptomatic pulmonarymetastasis by routine CR.
The study was approved by the institutional review board at the MassachusettsGeneral Hospital (MGH), Boston. Reports of diagnostic imaging studies aretransferred to an indexed text-search engine that allows real-time interactivesearches using Boolean operators (Folio Views; NextPage Inc, Lehi, Utah; systemsintegrator, Camberley Systems Inc, Needham, Mass). Patient and examinationidentifiers are included, and the "History," report "Body," and "Impression"are stored in separate fields. Reports were obtained for all CR examinationsfor which a history of melanoma had been provided during the years 1990 through1994. All patients would have at least 5 years of follow-up information atthe time of the search. Our clinic protocol recommends single baseline CRfor all thin (<1.0 mm) melanomas and annual CR for melanomas 1.0 mm orthicker. Since the care of our patients is often shared with outside dermatologistsand oncologists, not all patients receive an annual CR at MGH. All suspiciousCR findings from an outside hospital, however, are confirmed by a radiologicalexamination at MGH. Only chest radiographs that have been reviewed by theMGH Radiology Department were used in this study. A total of 1938 chest radiographsfrom 994 patients were identified.
All CR reports were reviewed, and the findings were classified as "negative,""positive," or "questionable." Reports with findings of masses, nodules, ill-definedopacities, hilar changes, or pleural changes were considered "positive." Reportswith "uncertain findings" and findings from follow-up tests requested by theradiologist were considered "questionable." For each patient with a "positive"or "questionable" CR report (n = 155) on the initial CR within the study period,we established the diagnosis and eventual outcome of the patient based ona manual search of hospital medical records, clinic medical records, the MGHtumor registry, and the Massachusetts State Death Registry. A database wascreated using Microsoft Access (Microsoft Inc, Redmond, Wash) to organize,combine, and assess information gained from record review.
Based on findings from medical record review, the 155 results that were"positive" or "questionable" were further categorized into 1 of the 5 followinggroups (Figure 1):
"Converted stage IV (C-IV)" included those patients(with localized or regional disease) who previously had no history of visceralmetastasis and were found to have pulmonary metastasis (stage IV disease)based on CR findings. All patients in this category had documented follow-upstudy results that confirmed that the finding on CR was a metastatic melanoma.
"Known stage IV (K-IV)" included those individualswho had a prior history of stage IV disease at a nonpulmonary site beforetheir first CR in this study.
"False positives" included those who had questionablefindings that were later shown not to represent pulmonary metastasis by additionalnoninvasive tests, which included repeated CR, computed tomography, magneticresonance imaging, bone scans, ultrasound examinations, or subsequent CR orbiopsy.
"Occult positives" were those persons whose positivefindings ultimately proved to be related to a cancer other than melanoma.In addition, since the pattern of metastases and follow-up regimen is differentfor mucosal and ocular melanomas, we excluded these patients (n = 6) fromour analysis as well.
"Nonevaluable" patients were lost to follow-updespite all attempts to identify an outcome.
All C-IV and K-IV patients were assessed for treatment of metastaticdisease and survival. Kaplan-Meier analysis was used to estimate the rateof survival among K-IV and C-IV patients, and the log-rank test was used tocalculate the P values between these 2 groups.
Of the 1938 chest radiographs from 994 patients with melanoma obtainedover the 5-year period (1990-1994), 1783 (92.0%) were normal and 155 (8.6%)were positive or questionable (Figure 1).On further analysis, of these 155 chest radiographs, 63 were eventually foundto be false positive (false-positive rate: 63/[1783 + 63] or 3.4%). Of theremaining 92 suspicious chest radiographs, 75 represented true pulmonary metastaseson follow-up and additional investigations. Each of the 75 chest radiographswere from separate individuals; 34 individuals had CR findings suggestiveof pulmonary metastasis in the context of known distant disease (K-IV), while41 (4% of all patients) demonstrated their stage IV disease through CR findings(C-IV). There were 7 chest radiographs that were later found to representcancers in the lung other than melanoma (1 squamous cell carcinoma of thelung, 5 adenocarcinomas of the lung, and 1 metastatic breast cancer), and6 suspicious chest radiographs were from persons with ocular or mucosal melanomaand were excluded from the study. Four chest radiographs were from patientswho had no further contact with MGH and thus whose outcome could not be determined.
For the 34 K-IV patients with known metastatic disease, 6 patients didnot receive further treatment at MGH and records of subsequent events wereunavailable. Of the remaining 28 patients who received further treatment,13 had surgical resection of their pulmonary disease in addition to othersystemic treatments. All 28 patients who had known treatments and 4 of 6 patientswho had unknown treatments eventually died; the status of the remaining 2patients with stage IV disease is unknown.
For the 41 C-IV patients whose chest radiographs demonstrated theirvisceral metastases, 3 patients were lost to follow-up and 7 patients electednot to receive any treatment. Of the remaining 31 treated patients, 10 underwentlung partial resection along with various systemic regimens. From this groupof 38 patients, 37 individuals died from their disease. One patient is stillalive more than 10 years after chemotherapy and thoracotomy for his pulmonaryrelapse. Three C-IV patients had unknown treatment regimens, and 2 of these3 persons died from metastatic disease; for 1 individual, no further informationon treatment or life status was found.
Figure 2 shows the Kaplan-Meiersurvival curves for C-IV and K-IV patients. There was no statistically significantdifference (P = .68, log-rank test) in survival betweenasymptomatic patients whose chest radiograph was the first indication of stageIV disease and patients whose chest radiograph revealed pulmonary pathologiccondition in addition to other known metastatic disease.
Although several studies have documented a low detection rate of pulmonarymetastases by CR in otherwise asymptomatic patients with melanoma, the practiceof routine surveillance CR still persists. Because the lungs represent a commonsite of metastasis for cutaneous melanoma, there is a rational, rather thanevidence-based, argument that early detection of pulmonary disease may enhancesurvival through earlier chemotherapeutic or surgical intervention. The utility of CR in detecting asymptomatic lung recurrencesdepends on the overall probability of first recurrence, the percentage ofsystemic recurrences, and the sensitivity and specificity of the procedure.The impact of CR on survival, however, depends largelyon effective treatments for pulmonary disease.
Over a 5-year period, approximately 8% of our study patients exhibitedCR findings that represented metastatic melanoma. For 41 individuals (4% ofpatients), routine CR represented the first indication of stage IV diseasein an otherwise asymptomatic individual. This rate is similar to rates reportedin the literature.15 In 34 cases, CR identifiedadditional metastatic deposits in patients already known to harbor distantdisease. Of the 155 positive/questionable radiographic findings, 63 (41%),eventually proved to be false positive (overall false-positive rate, 3.5%),although 7 additional unrelated pulmonary malignancies were diagnosed in theprocess. The high proportion of false positives may reflect, in part, an unusuallyhigh index of suspicion among radiologists, since a clinical history of melanomawas provided for all the retrieved requisitions. Moreover, our ascertainmentmay have enhanced sensitivity but diminished specificity because we includedall chest radiographs with any suspicious radiographic finding in the positive/questionablegroup that was further evaluated.
Several studies suggest that earlier detection of limited lung metastasesalong with aggressive surgical intervention may lead to an improved outcomefor patients with stage IV disease.2,6-8 However,these studies were based on a select group of patients with resectable pulmonarydisease. In one large series from the John Wayne Cancer Institute, Santa Monica,Calif, only 10% of all patients with metastatic melanoma involving the lungor thorax were eligible for surgical treatment.7 Westudied a general melanoma population and did not observe a survival advantagefor asymptomatic patients whose pulmonary relapse was initially detected byroutine CR. Our data suggest that, for most patients with melanoma, treatmentof those with pulmonary metastases even at the earliest possible time of detectiondoes not significantly contribute to overall survival. Since the patientswith stage IV disease were treated with a broad range of surgical and systemicapproaches, we did not attempt to further stratify these patients into treatmentgroups and assess outcome.
It is also possible that the observed benefit of aggressive surgicalresection reported in the literature is related to a more indolent tumor ora more aggressive immune response. It is thus interesting to note that the1 patient who had an ostensible cure of his stage IV disease (ie, >10-yearsurvival after relapse) and underwent a thoracotomy for his lung metastasisalso exhibited widespread melanoma-associated leukoderma—an immunologicalsign that has been associated with a better prognosis.16,17
Similar questions have been raised regarding follow-up tests for othercancers, including in colon and breast cancer survivors.18-21 Inseveral recent large trials involving breast cancer, frequent laboratory testsand radiological tests after primary treatment did not appear to improve survivaldespite an earlier diagnosis of recurrence.18,19 Comparabledetection rates (<10%) by CR in asymptomatic patients with colon cancerhave also been described.20
In summary, this study further challenges the use of CR in routine postmelanomasurveillance, since survival does not appear to be affected. Many studieshave already documented the questionable utility of large-scale CR screening,since most melanoma metastases are detected by simple history and physicalexamination and not by laboratory tests. The most definitive study, however,would be a multicenter prospective randomized trial of patients with high-riskmelanoma assigned to imaging vs nonimaging arms. However, until more effectivetreatments are developed, earlier identification of pulmonary relapse canpotentially benefit only a small, select group of patients with surgicallyresectable stage IV disease.
Corresponding author and reprints: Hensin Tsao, MD, PhD, Departmentof Dermatology, Massachusetts General Hospital, Bartlett 622, 48 Blossom St,Boston, MA 02114 (e-mail: firstname.lastname@example.org).
Accepted for publication May 28, 2003.
This study was partially supported by Career Development Awards fromthe American Cancer Society, Dermatology Foundation and American Skin Association(Dr Tsao); and the Jason Sabbag Melanoma Memorial Fund (Dr Sober).