Mixed and nonvaccine high risk HPV types are associated with higher mortality in Black women with cervical cancer

We studied the incidence of HPV genotypes in mostly Black women with cervical carcinoma and correlated histopathologic tumor characteristics, immune markers and clinical data with survival. Disease-free survival (DFS) and overall survival (OS) were recorded for 60 months post-diagnosis. Fifty four of the 60 (90%) patients were Black and 36 (60%) were < 55 years of age. Of the 40 patients with typeable HPV genotypes, 10 (25%) had 16/18 HPV genotypes, 30 (75%) had one of the non-16/18 HPV genotypes, and 20 (50%) had one of the 7 genotypes (35, 39, 51, 53, 56, 59 and 68) that are not included in the nonavalent vaccine. Mixed HPV infections (≥ 2 types) were found in 11/40 (27.5%) patients. Patients infected with non-16/18 genotypes, including the most common genotype, HPV 35, had significantly shorter DFS and OS. PD-L1 (p = 0.003), MMR expression (p = 0.01), clinical stage (p = 0.048), histologic grade (p = 0.015) and mixed HPV infection (p = 0.026) were independent predictors of DFS. A remarkably high proportion of cervical cancer cells in our patients expressed PD-L1 which opens the possibility of the use of immune checkpoint inhibitors to treat these cancers. Exclusion of the common HPV genotypes from the vaccine exacerbates mortality from cervical cancer in underserved Black patients.

genotypes included in the US Food and Drug Administration (FDA) approved HPV vaccine (Gardasil 9) has increased from 4 to 9, which has the potency to target 70% of the HPV infections. Black patients have a preponderance of non-16/18 HPV infections, including some rare genotypes with less known pathogenesis and outcomes. More studies on the distribution of HPV genotypes in racially and ethnically diverse patients and a better understanding of pathogenicity of these HPV genotypes are required for a more complete HPV vaccine for better coverage of the vulnerable populations.
The PD-1/PD-L1 axis is one of the immune-checkpoint pathways that have been found to help cancer cells evade the immune response 7,8 . PD-L1 expression has been studied for determining its therapeutic and prognostic role in cervical cancer. PD-L1 expression in cervical squamous cell carcinoma cells has ranged from 34.4 to 70.1% [9][10][11] . PD-L1 overexpression in tumor cells has been found to be associated with lymph node metastasis, vascular invasion, higher International Federation of Gynecology and Obstetrics (FIGO) stage and poor survival 9,11,12 . PD-1/PD-L1 inhibitors have been approved by the FDA for patients with recurrent or metastatic cancer unresponsive to chemotherapy. However, it can be a potential therapeutic option in patients with earlystage recurrent or locally advanced cervical cancer and who are unresponsive to chemotherapy 13 . Because of sparse data the efficacy of immune-checkpoint inhibitors has not been established as a first line therapy in primary cervical carcinoma, especially in Black patients, who commonly present with advanced stage disease.
Another biomarker for identifying patients for immunotherapy is the Mismatch Repair (MMR) or Microsatellite Instability (MSI) pathway which was recently approved by the FDA for use in different tumor types including cervical carcinoma 14,15 . Loss of function of one of the MMR proteins (MLH1, MSH2, MSH6, PMS2) leads to high rates of mutations that accumulate in repetitive nucleotide regions (microsatellites) inactivating the mismatch repair pathway. MMR-deficient tumors have 10-100 times more somatic mutations than MMR-proficient tumors leading to increased tumor antigen burden and immunogenicity. We speculate that MSI may have predictive and prognostic value in cervical cancer along with its co-expression with PD-1/PD-L1 and HPV genotype.
In this study we have attempted to characterize the clinicopathological features of cervical carcinoma in our patients and analyzed the expression of PD-1, PD-L1, MMR and HPV genotyping to explore their pathogenic and therapeutic potential.

Methods
Patient selection. Patients (n = 198) diagnosed with squamous cell carcinoma of the cervix over a 12-year period at a tertiary care hospital were identified. Sixty of these were found to have complete clinical data and adequate histologic specimens, and were grouped based on their race (Black and Other). Although the population we serve is heterogeneous and includes both African American and African Caribbean individuals, our electronic medical records based on a pre-determined classification permit self-reporting only as Black or African American. Histopathological specimens were reexamined to confirm the diagnosis. Paraffin blocks were retrieved, and tissue microarrays were prepared using two 3-mm punches. Clinical data including patient survival were recorded for up to 60 months post-diagnosis. The study was approved by the Institutional Review Board & Privacy Board (IRB) of the State University of New York, Downstate Health Sciences University, and patient consent was not required (IRB #1230970-3). The study methods complied with the local and federal government guidelines and regulations. The IRB determined that the study met the requirements for Waiver of Informed Consent as per the Office of Human Research Protections (OHRP) guidelines. This category includes materials previously collected for either non-research or research purposes. As this is a retrospective study, the clinical data and tissue samples included in the study were collected in the past and the patients were not available for consent at the time of the study.
Human papilloma virus (HPV) genotyping. Multiplex PCR and agarose gel electrophoresis were performed on DNA extracted from formalin-fixed paraffin-embedded tissue. To evaluate the DNA quantification after DNA extraction, we analyzed DNA measurement using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA). Thirty-three primers designed for high-risk HPV types (16,18,26,31,33,35,39,45,51,52,53,56,58,59,66,68,73,82) were used, and their unique specificity was confirmed by BLAST analysis (http:// www. ncbi. nlm. nih. gov/ BLAST/) (Qiagen, Germantown, MD) (Supplementary Information: Table ST 1). Primer selection for each reaction tube mix was based on a previous study 16 . A housekeeping gene (β-globin) common to all high-risk HPV types was utilized as internal control. Specimens that did not yield adequate DNA for HPV typing were retested. PCR products were analyzed by electrophoresis on a 2% agarose gel stained with ethidium bromide, band sizes were estimated by comparison with a 100 bp molecular weight marker (Invitrogen 100 bp DNA Ladder), and gels were photographed in a UV transilluminator (Thermo Fisher Scientific, Waltham, MA) (Fig. 1).
Immunophenotyping. Expression of PD-L1, PD-1, MMR proteins (MSH2, MSH6, PMS2 and MLH1) and p16 was characterized by immunohistochemistry, and the results were validated by using appropriate positive and negative controls. At least 100 viable tumor cells were evaluated for each of the proteins. Tumor viability was assessed by the presence of intact nuclear, cytoplasmic and membranous features, and devoid of any histologic sign of cell death (such as blebbing, cell shrinkage, nuclear fragmentation, chromatin condensation and chromosomal DNA fragmentation).
Immunophenotyping was performed by using a standard protocol starting with deparaffinization of tissue microarrays, antigen retrieval (60 °C; 12 min.), cell conditioning (64 min) and application of pre-primary antibody peroxidase. The specimens were then treated with the primary antibody (pre-diluted by manufacturer) ( www.nature.com/scientificreports/ Hematoxylin was used as counterstain (4-8 min), followed by bluing reagent (4 min). A matched negative control reagent was used for quality control. PD-L1 expression was evaluated by using SP263 rabbit monoclonal antibody and OptiView DAB IHC detection kit on Ventana BenchMark Ultra (Ventana, Roche Diagnostics, Oro Valley, AZ) following the manufacturer's protocol. Human term placenta was used as the positive control. Positive control staining was deemed acceptable if there was moderate to strong uniform membrane staining of trophoblast-lineage cells, and no staining of placental stroma and vasculature. PD-L1 expressing tumor cells (moderate to strong membranous staining) were quantified as negative (< 1% positive cells), low expression (> 1 to 49% positive cells), or high expression (> 50% positive cells) (Fig. 2).
PD-1 expression of tumor-infiltrating lymphocytes (TILs) was evaluated by using PD-1 monoclonal antibody and OptiView DAB IHC detection kit on Ventana BenchMark Ultra (Ventana, Roche Diagnostics, Oro Valley, AZ) following the manufacturer's protocol. Lymph nodes were used as positive control tissue. The percentage of immune cells with strong membranous staining under the light microscope was recorded. Strong membranous PD-1 expression in TILs was categorized as low or high based on values below or above the 50th percentile of PD-1 + TILs/mm 2 , respectively. MMR proteins were analyzed by using the Ventana MMR IHC monoclonal antibody panel for MSH2 (G219-1129), MSH6 (SP93), PMS2 (A16-4) and MLH1 (M1) and OptiView DAB IHC detection kit on Ventana Bench-Mark Ultra (Ventana, Roche Diagnostics, Oro Valley, AZ) following the manufacturer's protocol. The specimens were evaluated for the presence or absence of specific nuclear staining in the tumor cells. Colon cancer tissue with intact MMR expression was used as a positive control. An "intact" status was assigned to specimens with unequivocal nuclear staining in viable tumor cells. "Loss" of expression was assigned to complete loss of nuclear staining or focal weak or equivocal staining in the viable tumor cells in the presence of internal positive controls (lymphocytes, fibroblasts or normal epithelium) 17 .
p16 was evaluated by using CINtec histology antibody, anti-p16INK4a (E6H4) and OptiView DAB IHC detection kit on Ventana BenchMark Ultra (Ventana, Roche Diagnostics, Oro Valley, AZ). p16-expressing squamous cell carcinoma tissue was used a positive control. Immunohistochemical expression of p16 was defined as "positive" if there was strong diffuse nuclear positivity in tumor cells, while "negative" included those with patchy, wild-type p16 staining pattern 18 .

Statistical analysis.
Age was categorized using 55 years as cut-off for childbearing age. Clinical staging was determined after surgical management of cases and was based on International Federation of Gynecology and Obstetrics (FIGO) 2009 classification. Clinical stages I and II were combined into "early" clinical stage and stages III and IV as "late" clinical stage. Histologic grades were based on the extent of cellular differentiation, and categorized into well, moderate and poor. Patients were followed for up to 60 months after initial diagnosis. Disease-free survival (DFS) was computed based on the number of months from diagnosis to recurrence, and overall survival (OS) was the number of months from diagnosis to death. Cases with incomplete follow up were censored.
The results of the HPV PCR analysis were recorded as positive or negative (Fig. 3). Nine cases that tested positive only for the common high-risk HPV β-globin gene were treated as a separate group. Cases with specific type results were further subcategorized based on HPV genotypes 16 or 18 (16/18), non-16/18, mixed    Mixed HPV infections were defined as infection with > 1 HPV genotypes. Most common HPV genotypes (16, 18 and 35) were analyzed against all other HPV genotypes. When examining a specific HPV genotype, it includes all patients with that HPV genotype (including single and mixed infections). Patients categorized as "vaccine covered" had positive PCR results for HPV high-risk genotypes covered by the currently FDA-approved nonavalent vaccine (Gardasil 9), which includes high-risk types 16, 18, 31, 33, 45, 52, and 58. Correlations between different parameters were determined by Pearson Chi-Square or Fisher's exact test. All reported p values are two sided. A p < 0.05 was considered significant. Cumulative 60-month survival was calculated by the Kaplan-Meier method and analyzed by log rank test (SPSS, ver. 26).

Results
Patient characteristics. Sixty patients diagnosed with squamous cell carcinoma of the cervix were included in this study. All of the patients received standard treatment protocol as per the National Comprehensive Cancer Network (NCCN) guidelines based on clinical stage (FIGO classification). Treatment ranged from loop electrosurgical excision to radical hysterectomy with lymph node dissection. The mean age at diagnosis was 52.8 (range: 27-86) years. 54 (90%) patients were Black and 6 (10%) were Other (3 white; 3 others). 44 (73.3%) patients were diagnosed at an early clinical stage (I & II). About half of the patients' tumor histologic grade was poor (51.7%), followed by moderate (36.7%) and well (11.7%) (Table1).
Patients infected with HPV types covered by the currently available HPV vaccine had a higher recurrence and death (44.4% and 33.3%, respectively) as compared to those with non-vaccine covered HPV (29.4% and 23.5%, respectively).
More patients with negative HPV PCR had poor histologic grade (63.6% vs. 49%) as compared to patients with HPV PCR positive tumors. 2 of 10 (20%) with negative HPV PCR results had recurrence and subsequently died  www.nature.com/scientificreports/ of the disease. There was no significant difference in the prognosis with HPV PCR positive results. 6/9 patients (66.7%) with non-specific HPV PCR had late clinical stage cancer (p = 0.016). 5/55 (9.1%) patients lacked p16 expression in the tumor cells, and they were all Black. Loss of p16 correlated with higher recurrence (p = 0.022) and death (p = 0.003). Patients with negative p16 immunoreactivity showed loss of MMR (40%) (p = 0.011) and negative to weak PD-L1 expression (80%; p = 0.009), as compared to those with positive p16. 6/60 (10%) patients were Other and had favorable outcomes. All Other patients were diagnosed at an early clinical stage. Only one patient had recurrence and none died during the 60-month follow-up. Most of the Black patients had poor histologic tumor grade at diagnosis (28/55; 51.9%). More Black patients were diagnosed at late stage (16/55, 29.6%), and about a third of them (17/55; 35.4%) had recurrence, 14 of whom died subsequently (14/55; 29.2%). Younger women tend to have early stage cancer (80.6%) at diagnosis as compared to older patients (62.5%). At 5-year follow-up, the mortality rate among older patients (6/21; 28.6%) was slightly higher than that observed among younger patients (8/33; 24.2%).
Recurrence was more frequently observed in patients who presented at a late clinical stage (p = 0.028); more than half of the patients (57.1%) diagnosed in late clinical stage and only a quarter (25%) of patients in early clinical stage had recurrence. Similarly, greater number of patients diagnosed at a late clinical stage (42.9%) died as compared to those diagnosed at an early stage (20%). Patients with poor histologic grade tumors had higher recurrence (p = 0.022) and death (p = 0.046). No recurrence or death was noted in patients with well-differentiated tumors. Most of those patients who had recurrence (77.8%) subsequently died (p < 0.0001) ( Table 2).

Discussion
The results of this study confirm reported higher prevalence of non-16/18 HPV genotypes among Black women with cervical cancer. However, in contrast to published data, our patients with non-16/18 HPV types had worse prognosis. Those with mixed HPV type infection had even poorer outcome. Also, patients who tested negative for HPV by PCR and negative for p16 on immunohistochemistry (IHC) had poor survival. We also found high rate of PD-L1 expression in cervical cancer cells, which was surprisingly associated with better survival. Our findings suggest a unique tumor microenvironment in squamous cell carcinoma in Black women which could offer therapeutic opportunities leading to survival benefit.
Outside of the US a higher prevalence of non-16/18 HPV high-risk types have been reported than HPV 16/18 genotypes, and the former were found to be associated with a higher grade of cervical disease 22 . In Korea HPV 16 and 58 were found to be in almost equal frequency in a cervical cytology study of 1158 women; patients with HPV 58 had a significantly higher progression rate to high grade dysplasia 23 .
A meta-analysis of HPV types in invasive cervical cancer in Asia found that types 16, 18, 52 and 58 had similar prevalence 24 . Studies from the Middle East have also shown higher prevalence of non-16/18 HPV genotypes, with HPV 31 most prevalent in Yemen and HPV 33 in Kuwait followed by 35, 39, 45, 52 and 58 25,26 . These studies indicate that the prevalence of non-16/18 HPV types differs globally and their impact in different patient populations has been largely under-estimated.
Studies on HPV vaccine effectiveness have highlighted protection against HPV-related pre-cancers and cancers, with a recent meta-analysis estimating 51% reduction in high grade cervical dysplasia among 15-19-yearolds and 31% reduction among 20-24-year-olds 27,28 . In light of these encouraging results, the findings of our study are striking because the currently available 9-valent HPV vaccine (against types 6,11,16,18,31,33,45,52,58) does not include seven HPV types (35,39,51,53,56,59,68) that were detected in about one-third (33.3%) of our patients, mostly of Caribbean origin. Vaccines currently available to prevent the most common strains of HPV, may not offer the same level of protection in our Black population and other racial and ethnic groups globally. The diversity of patients and the prevalence of different HPV genotypes in the US and globally should be acknowledged as scientists guide further HPV vaccine development. To ensure successful cervical cancer prevention, physicians should continue to perform cervical cancer screening in women until complete high-risk HPV virus coverage is available in the vaccine.
Our patients with non-16/18 types had worse prognosis than those infected with 16/18 in early clinical stage. This finding fails to validate earlier studies which found that women with HPV 16/18 positive tumors had worse survival than those with other HPV types 21,29,30 .
One fifth of our patients had a mixed HPV infection and they were all found to have higher grade carcinomas (moderately to poorly differentiated) and a lower DFS and OS than those with a single HPV type infection. These tumors also had a higher PD-1 expression in TILs making them high-risk for aggressive course of disease although they could also prove to be targets for checkpoint inhibitor therapy.
Our patients with negative HPV PCR results and non-typeable HPV types had high tumor grade at diagnosis and were also diagnosed at a late clinical stage. These findings suggest that patients with non-typeable oncogenic HPV types should be treated with higher circumspection because of their aggressive behavior and limited knowledge about their genotype making them hard to include in the vaccines. Patients with negative p16 expression had significantly shorter DFS and OS as compared to those with p16-positive tumors. Although this observation is based on a small sample in our study, similar observations in other population studies have  36 . The p16 positivity is therefore considered a more sensitive indicator of HPV-mediated carcinogenesis in cervical cancers. Loss of the frequently targeted viral L1 gene after viral integration could be a reason for HPV-negative test results 37 , while other studies have identified methodological limitations (long-term storage, low tumor tissue) as probable causes 38 . PD-L1 expression in cervical cancer in our patients was 93% which far exceeds the reported frequency of 35-70% [9][10][11] . Expression of PD-L1 has been reported to be of variable prognostic information in neoplasms of different organs. It is a poor prognostic indicator in solid tumors, including esophageal, gastric, colorectal, and pulmonary adenocarcinoma, and a good prognostic indicator in lung squamous cell carcinoma 39,40 . Our study found PD-L1 to be a predictor of longer survival in early stage cervical carcinoma. Presence of higher PD-L1 increases the possibility of using targeted immunotherapy in these patients. Racial difference in PD-L1 expression of tumor cells had not been widely investigated. A recent single institution study of 114 cases, 36% of which were Black, concluded that PD-L1 expression in non-small cell lung cancer did not differ by race, clinical stage or smoking history 41 . Findings of a higher PD-L1 expression on cervical carcinoma cells in our population should be explored further in a larger multiracial study to identify the association of race with PD-L1 expression.
PD-1 expression in TILs surrounding cervical carcinoma showed no association with clinical stage, histologic grade, or HPV type, although PD-1 expressing TILs were greater in younger patients and patients with mixed and non-16/18 HPV infections. The prognostic significance of PD-1 expression in cervical carcinoma is still not fully elucidated, however, increase in PD-1 expression has been associated with poor prognosis in gastric, renal, nasopharyngeal and non-small cell lung carcinomas [42][43][44][45] , and increased survival in melanoma, glioblastoma, and ovarian, breast, and primary HPV-positive head and neck cancers [46][47][48][49][50] . Although not of prognostic value in our study, higher expression of PD-1 may be helpful to guide immunotherapy in the treatment of cervical carcinoma in patients with HPV infection of more severe types like non-16/18 and mixed, and patients with high stage carcinoma with limited treatment options.
MMR deficiency in solid tumors has recently been linked to susceptibility to immunotherapies targeting the PD-1/PD-L1 axis 14,15 . Loss of MMR proteins has been shown to correlate with tumoral PD-L1 expression in colorectal and endometrial carcinomas, but this association has not been examined in cervical carcinoma, where MMR deficiency is less common. In the current study 8.3% (5/60) patients were MMR-deficient whereas other reports have documented 25.8% (17/66) cervical carcinoma patients with MMR deficiency and significant correlation with PD-L1 expression on tumor cells 51 . All of our five MMR-deficient patients expressed PD-L1.
Race, specifically the Black race, has been identified as one of the most common demographic factors associated with cervical squamous cell carcinomas 21 . Blacks have worse survival rates likely due to more advanced-stage tumors at diagnosis and less opportunity likely to receive adequate treatment as compared with white women [52][53][54] . This trend was similarly reflected in our study where more than half of the Black patients had poor histologic tumor grade at diagnosis, about a third were diagnosed at late clinical stage and a third had recurrence who subsequently died. On the other hand, all non-Black patients, albeit in small number, had favorable prognosis; all of them were diagnosed at an early stage and all survived.

Conclusions
Most Black patients in our study were infected with non-16/18 HPV genotypes, which were associated with a lower disease-free and overall survival. HPV 35 was the most common type isolated in our patient population, and this HPV genotype is not included in the commercially available nonavalent vaccine. Exclusion of HPV 35 and other common genotypes (39,51,53,56,59,68) from the vaccine exacerbates mortality from cervical cancer in these underserved communities. Patients with HPV 35 and those with mixed HPV infection had shorter DFS and OS, especially in early clinical stages. A remarkably high percentage of PD-L1 expression in cervical cancer cells was observed in our predominantly Black patient population and a high PD-1 expression on TILs in younger patients. These findings suggest that immune checkpoint inhibitors could be given consideration in the management of primary invasive squamous cell carcinoma of the cervix among Black patients.