Background and design: Cutaneous T-cell lymphoma (CTCL) frequently presents a difficult diagnostic challenge for the clinician and pathologist. To assess the diagnostic validity of conventional histopathologic findings in CTCL, pretreatment skin biopsy specimens were scored prospectively and independently by a panel of five to seven dermatopathologists and pathologists. Scores were compared with disease outcome. Repeatability of these scores was examined among observers and for the same observer. The study population consisted of 165 subjects, initially referred for suspected mycosis fungoides or Sézary syndrome. Ninety-two patients determined to have CTCL have been followed up for 6.3 +/- 3.5 years (mean +/- SD) and are categorized according to disease outcome: 22 are in complete remission, 35 are in partial remission, three have progressive lymphoma, 15 died of disease, 13 died of other causes, and four were unavailable for follow-up. Seventy-three patients determined not to have CTCL have been followed up for 5.3 +/- 3.2 years without subsequent clinicopathologic evidence of CTCL. These longitudinal data allowed comparisons of the clinical course with the original histologic interpretations.
Results: Data showed that the histologic scores rendered by the pathology panel did not correlate with stage of disease and were not an accurate predictor of clinical outcome, because the histologic ratings did not discriminate between patients who eventually had complete remission and those with either progressive lymphoma or who have died of disease. The results also substantiate the low inherent reliability of histopathologic findings in CTCL. Large differences existed among pathologists in scoring the study populations and repeated reading of selected cases by the same panel member resulted in a change of diagnosis 15% of the time. Among the histologic features evaluated, only the presence of mitoses in the infiltrating cells showed a trend toward an unfavorable outcome.
Conclusion: Pathologic diagnosis in the CTCL disease spectrum should be interpreted with caution and then only in conjunction with the clinical evaluation. As expected, the use of an average value from a panel of readers added a component of stability to the histologic interpretation.