Real Time PCR analyses for MCPyV
Serial dilutions of genomic DNA from a fresh frozen MCC tissue sample associated with the MCV350 strain were employed to calibrate the real time PCR method. The reference curve, ranging 0.1–10 ng genomic DNA, corresponded to 10–1000 cells (as determined by simultaneous amplification of the human RNaseP gene). MCPyV (confirmed by sequencing) was readily detected by this assay in fresh frozen MCC tissue but in only 1/52 non-MCC controls (DNA samples from 12 colon cancers, 20 breast cancers, and 20 normal peripheral blood lymphocytes).
MCPyV DNA was detected in 28/35 MCC samples, including 17 (74%) of 23 MCC individual cases. The range of viral abundance (amount of viral DNA/amount of cellular DNA) in these samples was 0– 1.2 with a mean and median viral abundance of 0.1275 and 0.001, respectively. ()
Viral abundance of Merkel cell polyoma virus (MCPyV) and expression of selected tumor proteins in 23 Merkel cell carcinoma cases.
MCPyV DNA in the positive samples presented a discontinuous association of MCPyV abundance (). Two distinct sub-groups of MCPyV positive tumors were discerned. Nine MCC samples had higher levels of MCPyV DNA (median 0.145 copy/cell) than 8 samples with very low levels of MCPyV DNA (median 0.001). No MCPyV DNA was detected in the remaining 6 MCC samples.
Figure 1 Distribution of MCPyV DNA level in virus positive MCC tumors. Copies of MCPyV DNA, standardized to copies of a cellular gene (RNAse P) are plotted on a logarithmic scale. Analytic sensitivity was 0.0005 viral copies/cell. Six MCC tumors were below this (more ...)
Concordance of viral-abundance was very high (coefficient of variation 0.11–0.26) in blinded duplicate samples from four patients (). MCPyV DNA was also detected in 1/26 uninvolved skin samples from MCC patients, with very low viral abundance (0.0003 copy/cell) (data not shown).
Concordance of viral copy number in same patients
Tissue Biomarkers (IHC)
All tumor tissues in this study met criteria for MCC diagnosis. Classic morphologic types were represented, and all were TTF-1 and CD99-013/CD 99-mic 2 negative. Staining for neuron-specific enolase, synaptophysin and most cytokeratins (CAM 5.2, MAK-6 and AE 1/3) were positive in all samples. Some biomarkers were associated with variable positive and negative results; these are shown in arranged by decreasing viral abundance. Chromogranin expression was positive or strongly positive in all tumors with higher viral abundance, including 18/23 positive overall. Epithelial markers EMA/BER EP4 stained positive in 20/23 cases. Similarly, the neural cell adhesion molecule CD56 stained positively in 19/23. CK 20 was uniformly positive for perinuclear punctuate staining in all but two cases with lower viral abundance. CK7 was positive only in some of the low/no viral abundance tumors.
Two MCC subgroups were readily distinguished by pRb expression; 9 positive, 14 negative ( and ). pRb-positive MCC tumors had relatively high levels of MCPyV DNA (median 0.145 copy/cell); MCPyV DNA was detected at very low levels (n=8, median 0.001) or not detected (n=6) in the pRb-negative tumors. TdT expression was observed in 5 of 8 pRb-positive tumors but in none of the pRb negative tumors. MCPyV viral-abundance was significantly associated with pRb and TdT expression (P<0.0001 and P=0.003, respectively, by Kruskal-Wallis). Higher viral abundance, pRb-expressing tumors were associated with higher CD44 expression (Wilcoxon P=0.02), absent CK7 expression (P=0.05), and lower p53 expression (P=0.06). Viral copy number was not related to the quantity (size) of tumor tissue in the DNA/PCR sample (Spearman R=−0.20, P=0.24).
Illustrative examples of correlation between immunohistochemical expression of pRb and p53 in MCC with high and low viral abundance.
We investigated whether selected MCC clinical characteristics differed between the two MCC subgroups (). The higher viral abundance, pRb-expressing subgroup tended to be younger (P=0.07) but did not differ by sex (P=0.64). MCC primary site, cell type, and presence of positive lymph nodes (either sentinel nodes or nodes in regional lymph node dissections) also did not differ by subgroup. Distant metastases occurred in 22% of patients with higher abundance, pRb-expressing tumors, compared to 43% of the other patients (P=0.4). Corresponding median survival times were 86 months and 20 months, respectively (Kaplan-Meier log-rank P=0.015). Adjusted for age, the relative hazard for death was 0.15 (95% confidence interval 0.02–1.24, P=0.08) with higher abundance, pRb-expressing tumors.
Viral abundance of Merkel cell polyoma virus (MCPyV) and patient and clinical characteristics
These data confirm that MCPyV is associated with most, but probably not obligatory for, MCC (1
). Like others (1
), we found that 24% of MCC had no detectable MCPyV DNA. An additional 40% had <1 viral copy/300 cells. Did MCC tumors with no or low viral abundance lose the MCPyV genome, as would occur with “hit and run” carcinogenesis? We think not, because such loss would result in uniform distribution of viral abundance, whereas our samples had either ≥1 viral copy per 16 cells or <1 viral copy per 300 cells. One should also consider whether the viral abundance in the higher category (0.06 – 1.2 viral copies per cell) would be sufficient to contribute to neoplasia. The true number of infected cells in a tumor cannot be determined with our methods. However, viral levels were highly concordant in our blinded duplicates of metachronous and recurrent tumors, suggesting that admixture by variable numbers of non-MCC cells in the tissue specimens played little role in the observed variation in viral abundance. Malignant transformation certainly could arise from clonal integration of the MCPyV genome and expression of its large T antigen in MCC tumors, events for which there is good evidence (1
). In addition, it is plausible that an infected cell could contribute to transformation of uninfected neighboring cells by paracrine mechanisms, as suggested for Kaposi sarcoma-associated herpes virus (19
Finally, we could identify no differences in the clinical characteristics or protein expression patterns between MCC with low-abundance virus and MCC with undetectable virus ( and ). This similarity was our rationale for considering these low/no virus MCC to be one subgroup. It also suggests that MCPyV may be detected in bystander cells, in which case the virus may be incidental to the malignancy.
Thus, our data support two virologically distinct MCC subgroups. Higher viral abundance is characterized by immunohistochemical detection of pRb and TdT, as well as a tendency for improved patient survival. MCC tumors with very low or undetectable levels of MCPyV DNA lack detectable pRb and TdT expression, and they have a tendency for poorer survival. These observations suggest that higher viral-abundance MCC may depend upon a pRb-mediated oncogenic pathway. In contrast other MCC, with few or no detectable copies of MCPyV, may depend on alternate oncogenic pathways such as p53. Future, larger studies are needed to corroborate our findings and should also consider the effects of immune deficiencies, ultraviolet radiation exposure (23
), tumor stage, and therapeutic interventions and responses.