J. M. Obeid1, G. Erdag3, T. Bullock4, N. Wages2, C. L. Slingluff1 1University Of Virginia,Surgery,Charlottesville, VA, USA 2University Of Virginia,Public Health Sciences,Charlottesville, VA, USA 3Johns Hopkins University School Of Medicine,Dermatology,Baltimore, MD, USA 4University Of Virginia,Pathology, Research,Charlottesville, VA, USA
Introduction: Infiltration of melanoma metastases by CD8+ T cells predicts improved survival and is believed to reflect immune-mediated tumor rejection. Thus, CD8 count is used as a marker of prognosis and response to immune therapy in clinical trials. Increases in CD8 T cells between pre- and post-treatment biopsies may reflect response to immune therapy, which is increasingly important for assessing effects of combination immunotherapies. Tumor heterogeneity may complicate these measures, but there are insufficient data to address heterogeneity when tracking changes in CD8 infiltrates. Furthermore, there is a need to assess differences in immune infiltration between synchronous and subsequent tumors. We hypothesized that variation of CD8 T cell counts among different samples of the same or synchronous metastases would be limited to a coefficient of variation (CV) of less than 50% of the mean. We also hypothesized that CD8 counts would decrease over time between metachronous tumors.
Methods: Tissue microarrays (TMAs) were constructed from 197 melanoma formalin-fixed paraffin-embedded tumors from 154 patients, of which 27 had 2 or more tumors resected at different times, and six patients had 2 tumors resected simultaneously. For each tumor, up to four 1 mm diameter tissue cores were included in the TMA. The number of CD8 T cells per core was determined by immunohistochemistry. Mean, standard deviation (SD), and coefficient of variation (CV = SD/mean) were calculated for tumors with 3-4 evaluable cores (N=175). In patients with metachronous tumors, CD8 counts of the first and second tumors were compared with a paired T-test. For simultaneous metastases, CD8 counts were studied for differences in decile ranks among all 197 tumors.
Results:CD8 counts varied widely among different cores of the same tumors (average CV 0.77, 95% CI: 0.70 to 0.84). The CV was greater for lower means (CV>0.7 for mean <134 CD8/mm2, CV<0.5 for >294 CD8/mm2). The inverse association of CV with the mean was significant (r = -0.38, p<0.0001). Among the 6 patients with simultaneous excision of two tumors, 4 pairs of tumors had counts in the same decile, 1 differed by 1 decile and 1 by 2 deciles. Among the 27 patients with metachronous tumors, if the first tumor had CD8 counts higher than the median (84 cells/mm2), CD8 counts decreased by 47% in the second tumor (p=0.005). For those with CD8 counts lower than the median, CD8 counts trended higher in the second tumor (p=0.058, +140%).
Conclusion:Heterogeneity among tumor samples was greater than hypothesized, but means across 3-4 cores were similar between synchronous metastases. In patients, differences in CD8 T cell counts after treatment may be explained by heterogeneity if tumor samples are small and especially if differences are less than 2-fold. The impact of tumor heterogeneity on CD8 T cell counts may be minimized by taking multiple samples of each tumor and powering clinical trials to allow for heterogeneity.