Lifetime Exposure to Cigarette Smoke, B-Cell Tumor Immune Infiltration, and Immunoglobulin Abundance in Ovarian Tumors.

We included women from the NHS and NHSII who reported on parental cigarette smoke exposure; 45% reported ever smoking in adulthood ( Table 1 ) and 66% reported at least one parent smoked in the home ( Supplemental Table 2 ). During follow-up (NHS: 1982–2016 NHSII: 1999–2017), 1,448 incident ovarian cancer cases were identified and tissue was obtained and included in the TMAs for 530 cases ( Supplemental Figure 1 ). Parental smoke exposure data were available for 395 cases and of those, successful image analysis was completed for 327 cases on the first panel and 383 cases on the second panel. Cases in this analysis were similar to all cancer cases, although cases in this study were slightly more likely to be diagnosed at an earlier stage and later in follow-up ( Supplemental Table 3 ) ( 6 ). At baseline, women with adulthood cigarette smoke exposure were slightly more likely to use oral contraceptives and more likely to have parental cigarette smoke exposure compared to women without adulthood smoke exposure ( Table 1 ). The percent positive for each immune marker, except IgA, was higher in women with Type II tumors compared to Type I tumors ( Supplemental Table 4 ). Among cases, women who ever versus never smoked in adulthood had higher odds of having IgA in the tumor (OR: 1.54, 95%CI: 1.14, 2.07; Table 2 ). Similarly, women who ever smoked cigarettes in adulthood with no parental cigarette smoke exposure versus never smoked with no parental cigarette smoke exposure had an increased odds of IgA in the tumor (OR: 2.03, 95%CI: 1.18, 3.49). Both findings were statistically significant after Bonferroni correction. Results were similar for IgG in the tumor, although these findings were not statistically significant (OR: 1.27, 95%CI: 1.00, 1.62 and OR:1.42, 95%CI: 0.92, 2.21, respectively). These findings were similar among Type II tumors and the NHS cohort ( Table 3 ; Supplemental Table 5 ). When we explored former or current smoking separately, findings for IgA were stronger for former (OR: 1.62, 95%CI: 1.18, 2.21) than current smokers (OR: 1.24, 95%CI: 0.74, 2.08) compared to never smokers ( Supplemental Table 6 ); although this may be due to the small proportion of women who were current smokers at diagnosis of ovarian cancer. When stratifying tumors by levels of intratumoral CD3 + T cells, we observed women who ever versus never smoked in adulthood had higher odds of IgA in the tumor (OR: 1.76, 95%CI: 1.16, 2.67) when CD3 + T cells were above the median but not below the median (comparable OR: 1.15, 95%CI: 0.74, 1.74) ( Table 4 ). Statistically significant patterns were observed for women who ever smoked in adulthood and did not have parental smoking exposure compared to women who never smoked and had no parental exposure (high CD3 + : OR: 2.65, 95%CI: 1.31, 5.36; low CD3 + : OR:1.37, 95%CI: 0.59, 3.22). We did not observe an association between parental or adulthood cigarette smoke exposure with risk of ovarian cancer by high versus low total B cells, naïve and memory B cells, plasma cells, or B-regulatory cells ( Supplemental Table 7 ). Additionally, we did not observe any association with parental or adulthood cigarette smoke exposure with risk of ovarian cancer classified by high IgA on B cells, or IgM ( Supplemental Table 8 ). However, women with parental cigarette smoke exposure versus none had a higher risk of developing ovarian tumors with low IgG (HR: 1.51, 95%CI: 1.10, 2.09), although the p heterogenity of 0.02 did not meet the Bonferroni corrected significance. Women who ever versus never smoked cigarettes during adulthood had a lower risk of developing ovarian tumors with low levels of IgA and IgG (HR: 0.71, 95%CI: 0.53, 0.96, p heterogenity =0.03 and HR: 0.74, 95%CI: 0.56, 0.99, p heterogenity =0.05, respectively). However, neither association was statistically significant at the Bonferroni corrected level. Similar, albeit weaker, associations were observed among Type II tumors ( Supplemental Table 9 , Supplemental Table 10 ).

We included the Nurses’ Health Study (NHS) and NHSII, two prospective cohorts. The NHS enrolled 121,700 U.S. female registered nurses in 1976 aged 30–55 and the NHSII enrolled 116,429 female registered nurses aged 25–42 in 1989 ( 24 , 25 ). The NHS and NHSII followed participants with biennial questionnaires including questions regarding early life exposures, adult health behaviors, and medical history. The study protocol was approved by the institutional review boards of the Brigham and Women’s Hospital, Harvard T.H. Chan School of Public Health, Advarra Institution Review board (IRB #00000971), and those of participating registries as required. Participants provided implied consent for the questionnaire and signed a medical release form for acquisition of tissue and medical records. This study was conducted in accordance with the Belmont Report and U.S. Common Rule. Assessment of parental smoking and adult exposure to cigarette smoke has been described previously ( 6 ). Briefly, in 1982 (NHS) and 1999 (NHSII), participants were asked if their parents smoked while living with them, or inside the home. If a participant selected her mother, father, or both, she was considered to have parental smoke exposure. For both cohorts, at baseline and each subsequent questionnaire, participants reported their current and past smoking status. Those responses were categorized into never or ever smoked cigarettes, as past and current smoking prior to diagnosis have both been associated with poor outcomes among women with ovarian cancer ( 5 ). To evaluate lifetime exposure to cigarette smoke, a composite variable was created using parental smoke exposure and adult smoking (never smokers with no parental smoke exposure, never smokers with parental smoke exposure, ever smokers [including former and current] without parental smoke exposure, and ever smokers with parental smoke exposure). Participants self-reported ovarian cancer diagnoses via questionnaire or were identified via linkage to the National Death Index ( 26 , 27 ). Medical record review, including pathology reports or linkage with state cancer registries were used to confirm cases and a gynecologic pathologist (JH) reviewed pathology reports to obtain stage. Measurement of immune markers have been described previously ( 6 ). Briefly, tumor tissue was obtained from 530 ovarian cancer cases and seven tissue microarrays (TMAs) with three 0.6-mm cores per case were created. A gynecologic pathologist (JH) identified areas of tumor on the slides and evaluated histology and grade. TMAs were stained using AKOYA Biosciences OPAL™ 7-Color Automation immunohistochemistry kit (Waltham, MA) and analyzed for B cells and Ig on two panels ( Supplemental Table 1 ). The slides were imaged using Vectra ® 3 Automated Quantitative Pathology Imaging System and were quantitatively analyzed using HALO software (Indica Labs, New Mexico) for positive cells with staining in the cytoplasm or nucleus, according to visual intensity for each marker. Both panels included pan-cytokeratin to differentiate tumor from stroma. We considered immune cells that were infiltrating into areas of tumor given that TMA cores were selected to maximize tumor area and immune cell infiltration in the tumor compartment has been most strongly associated with ovarian cancer-related outcomes ( 28 ). We excluded 88 cases from the first panel and 21 from the second panel without staining results due to damage or folding of the tissue cores. Our analysis evaluated parental smoking and adult smoking with counts of cells positive for each marker and percentage (number of positively stained tumor cells in the core divided by the total number of tumor cells in each core, times 100) of total B cells (CD19 + ), naïve and memory B cells (CD20 + ), plasma cells (CD138 + ), B-regulatory cells (CD19 + TNFR2 + ), IgA on a B cell (CD19 + IgA + ), IgA (IgA + ), IgG (IgG + ), and IgM (IgM + ) within the tumor compartment. We excluded women with a bilateral oophorectomy (before ovarian cancer diagnosis; n=10,840), missing date of birth (n=476), or menopause due to radiation (n=221) as well as women who did not answer the question on parental smoking exposure or who died before assessment of parental cigarette smoke exposure (1982 for NHS and 1999 for NHSII; n=50,429). We additionally excluded cases not included on the TMAs or that were missing immune cell data (n=1,051) for a total cohort size of 165,691 women including 395 ovarian cancer cases. Selected blocks contained primary epithelial ovarian or peritoneal tumor tissue, and to our knowledge, all cases were treatment naïve. To evaluate the measured markers continuously, we conducted a case-only analyses using beta-binomial models to estimate odds ratios (OR) and 95% CIs, interpreted as the ratio of odds that a cell was positive for the marker(s) of interest in the exposed versus unexposed groups ( 29 ). To address multiple cores per case, participant was included as a random effect; models were adjusted for fixed effects of: age at cancer diagnosis, year of cancer diagnosis, cohort (NHS, NHSII), histotype (Type I [low grade serous, endometrioid, clear cell, mucinous] versus Type II [high grade serous, poorly differentiated, carcinosarcoma, transitional/Brenner]), stage at diagnosis (I, II, III, IV), and adjusted for known ovarian cancer risk factors, including menopausal status (premenopausal, postmenopausal), hormone therapy use (ever, never), oral contraceptive use (never, <1, 1–5, 5–10, 10+ years), parity/breastfeeding (nulliparous, ≥1 child/no breastfeeding, ≥1 child/breastfed), hysterectomy (not related to ovarian cancer diagnosis: yes, no), tubal ligation (yes, no), family history of breast or ovarian cancer (yes, no), and body mass index (BMI; <20, 20–24.9, 25–29.9, ≥30 kg/m 2 , missing), updated by self-reported questionnaire every two to four years. We also adjusted for CD3 + cells (tertiles) to identify the independent effect of B cells, given that we previously observed an association between smoking and T cell infiltration ( 6 ). We conducted secondary analyses among Type II tumors only, NHS only, and stratified by low/high CD3 + cells in the tumor (based on the median). Given the multiple immune cells being evaluated, we calculated the number of effective independent tests to be 5 using the method of Li et al. ( 30 ), which takes into account correlations between the markers, resulting in a Bonferroni corrected P -value of 0.01 (0.05 / 5) to be considered statistically significant ( 30 , 31 ). We conducted competing risks Cox proportional hazards models to evaluate hazard ratios (HR) and 95% confidence intervals (CIs) of smoking exposure with risk of ovarian cancer by each immune cell type listed above, dichotomized at the marker-specific median of percent positivity. The average percent of cells in a core positive for a marker across all cores for each case was calculated and defined as being above or below the population median. Models were stratified by age, calendar year of questionnaire, and cohort (NHS, NHSII) and adjusted for the ovarian cancer risk factors listed above. Heterogeneity between the HRs for ovarian cancer with immune cell percentage below the median versus greater than or equal to the median was evaluated using the likelihood ratio test ( 32 ). We conducted secondary analyses restricted to Type II ovarian tumors ( 33 ). Analyses were conducted in SAS version 9.4 (SAS Institute, Cary, NC, USA) and R version 4.1.0 (R Core Team, Vienna, Austria). The data generated in this study are available upon request from the corresponding author.

Overall, in this large study, we observed that patients with ovarian cancer who ever versus never smoked cigarettes in adulthood had increased odds of having intratumoral IgA, which was stronger for women who had no parental cigarette smoke exposure or in tumors with higher CD3 + T cell tumor infiltration. We did not observe significant associations between parental or adulthood cigarette smoke exposure and overall B cell abundance or specific types of B cells, even though plasma cells (a cell type that originates from B cells) are responsible for Ig production. Additionally, women with versus without parental cigarette smoke exposure had a suggestively higher risk of developing ovarian cancer with low IgG, while ever versus never adult smokers had a suggestively lower risk. Associations were similar for Type II tumors. Previous studies have consistently observed an association between prediagnosis smoking and increased ovarian cancer mortality ( 5 , 34 , 35 ), although histotype-specific associations, which are often challenging to study due to limited case numbers, have varied. A pooled analysis of 19 case-control studies observed higher mortality among both current and former smokers (HRs 1.17 and 1.10, respectively) and, in analyses by histotype, current smoking was associated with worse survival among those with mucinous or serous tumors ( 34 ). Similarly, in the NHS/NHSII, we previously observed increased risks of ovarian cancer-specific mortality among former and current smokers (HRs: 1.19–1.21) and those with ≥20 vs. <20 smoking pack-years (HR: 1.28); a significant association with pack-years was observed for high-grade serous tumors only ( 5 ). In the Million Women Study, current, but not former, smoking was associated with worse ovarian cancer survival (HR: 1.17), but no significant association with current smoking was observed when stratified by tumor histotype, due to limited power ( 35 ). Despite multiple studies demonstrating that nicotine can bind to the nicotinic receptor on circulating B cells, inhibiting their development and function ( 14 ), we did not observe any associations of smoke exposure with B cell infiltration in ovarian tumors. This may be due to the differing contexts of systemic and anti-tumor immunity ( 36 ). Further, it is possible that any such effect could wane with time and only 11.8% of ovarian cancer cases smoked near the time of diagnosis. The lack of association for parental smoke exposure with B cell infiltration is consistent with animal model studies of smoke or nicotine exposure in adolescent rats that observed no changes in mitogenic response of B cells compared to control ( 37 , 38 ). Several ( 14 , 16 , 17 ), but not all ( 19 , 20 ), prior studies reported increased IgA in serum and saliva of adult smokers versus non-smokers. Similarly, we found an increased odds of IgA presence in ovarian tumors in ever smokers, which was stronger for smokers without parental smoking exposure. This suggests that early exposure could reduce IgA production in response to adult smoking. Interestingly, one study of postnatal smoke exposure in rats observed significantly reduced antibody forming cell response compared to control ( 38 ). We also observed that ever smoking was more strongly associated with IgA positivity in the tumors among those cases who also had high CD3 + T cells. Notably, prior studies have shown that T cells were only predictive of improved survival when Ig-generating plasma B cells were present ( 9 ). A protective role for IgA responses has also been identified in endometrial carcinoma ( 11 ), while non-small cell lung cancers with denser IgA infiltration show stronger accumulation of CD3 + T cells ( 12 ). Although further study is needed, our data suggest that ovarian carcinomas associated with smoking could be more immunogenic, albeit more aggressive. Similarly, smoking-associated lung tumors have been associated with higher tumor immunogenicity and stronger immune-environments ( 39 ). Our findings suggest that changes in adaptive immunity likely do not explain why pre-diagnosis adult cigarette smoke exposure is associated with worse survival in women with ovarian cancer ( 4 , 5 ). Further, IgA enhances tumor immune responses through coordination of T and B cells and was related to improved survival when IgA colocalized with polymeric Ig receptor (pIgR) ( 9 ). pIgR binds polymeric immunoglobulins and transports dimeric IgA from the basolateral surface of epithelial cells to the apical side of the mucosa. As it approaches the luminal surface, pIgR undergoes proteolytic cleavage, releasing secretory IgA. pIgR has been identified in normal fallopian tube, ovarian, and omental tissues ( 9 ), and in human ovarian and endometrial cancer cells ( 9 , 40 ). The recognition of IgA by pIgR prompts directional transcytosis of dimeric IgA through these tumor cells ( 11 , 12 , 41 ). Moreover, IgA:pIgR co-localization correlates with enhanced outcomes in human ovarian cancer. Notably, McGrath et al. observed that cigarette smoke attenuates the upregulation of pIgR in nasal mucosa ( 42 ). It is possible that smoking may reduce pIgR and disrupt the IgA-pIgR complex, leading to worse outcomes. However, more work is needed to fully elucidate the biological mechanisms underlying our observations. Studies of the impact of adulthood cigarette smoking on serum and saliva IgG levels have observed higher ( 16 , 17 ) and lower IgG levels ( 14 , 15 , 18 , 20 ). Within ovarian tumors, we observed an increased odds of IgG positivity among ever versus never adult smokers and a decreased risk of developing ovarian cancer with low levels of IgG among adult smokers. Conversely, there was an increased risk of developing ovarian cancer with low levels of IgG among those who were exposed to cigarette smoke early in life, though none of these findings met statistical significance level. Chronic tobacco smoke increases DNA damage ( 43 ), which can lead to neoantigen presentation ( 44 ) thus stimulating an antibody response in the tumor ( 39 ); IgG may also be upregulated to neutralize toxins found in cigarette smoke ( 17 ). Further, exposure to chemicals in cigarettes has been associated with B cells shifting from IgG to IgE production ( 45 ). It is possible that parental cigarette smoke exposure could alter overall Ig production even into adulthood by reducing antigen response ( 46 ). Further research is needed to validate our findings. This study was the first to evaluate exposure to cigarette smoke over the life course with ovarian tumor infiltrating B cells and Igs. We utilized two large prospective cohort studies that allowed for controlling of potential confounders and used validated multiplex immunofluorescence assays. However, this study was limited in the ability to evaluate individual histotypes, to assess subpopulations of IgG (i.e., IgG1, IgG2, IgG3, IgG4) that may have different associations with smoking ( 47 ), and to distinguish if Igs were bound to an immune cell or tumor cell. We also used tissue microarrays, which may not capture heterogeneity across the tumor. Our study consisted of primarily non-Hispanic white women with fairly high levels of education, which may limit the generalizability. We used self-reported cigarette use prospectively collected over time and retrospective report of parental cigarette smoke exposure, which may have misclassification due to recall bias (early life) or social desirability bias (adult use); however, prior work in the NHSII cohort has demonstrated high concordance between the participant’s report of mother smoking prenatally and the mother’s self-report ( 48 ). Overall, our study did not observe strong associations between smoking across the lifecourse with B cell infiltration in the ovarian tumor immune microenvironment. However, adult smoking may lead to increased IgA and IgG abundance in ovarian tumors, while parental smoking exposure may increase risk of developing ovarian tumors with low levels of IgG. These findings may indicate a priming effect of Ig switching when exposed to cigarette smoke early in life, and an overactive antibody response when exposed to cigarette smoke in adulthood. Future work is needed to evaluate whether smoking leads to worse survival through alterations in anti-tumor immunity and how smoking may impact the functionality of Igs within the tumor microenvironment.

Smoking cigarettes in adulthood is associated with a modest increased risk of ovarian cancer, primarily attributed to increased risk of developing mucinous tumors ( 1 – 3 ). In addition, smoking cigarettes in adulthood and exposure to parental smoking early in life have been associated with ovarian cancer outcomes. Notably, both former and current smoking before diagnosis were associated with poor ovarian cancer-specific survival compared to never smokers ( 4 , 5 ). While the underlying biologic mechanisms linking cigarette smoke exposure with worse ovarian cancer survival have not been established, one hypothesis is that pre-diagnosis smoking impacts the anti-tumor immune response. We previously reported that childhood and adolescent exposure to cigarette smoke in the home (versus none) was associated with a 45–50% increased risk of developing ovarian cancer with low intratumoral levels of total CD3 + T cells and recently activated cytotoxic T cells (CD3 + CD8 + CD69 + ) ( 6 ). Cytotoxic T cells trigger tumor apoptosis by secreting toxic substances (e.g., granzymes) into tumor cells and damaging tumor cell DNA via the Fas/Fas ligand pathway ( 7 ). The B cell response is another important factor in ovarian cancer outcomes. For example, a study of human high grade serous ovarian cancer omental metastases found that intratumoral B cells produce cytokines and chemokines, such as GM-CSF, interferon gamma, interleukin 12, CXCL10, and CXCL8, which recruit antitumor immune cells (e.g., macrophages, dendritic cells, T cells, natural killer cells) ( 8 ). Also, in vitro studies suggested that tumor-derived IgG, the predominant immunoglobulin in omental metastases, recognizes and binds to tumor antigens, forming a complex that may activate antigen-presenting cells ( 8 ). Further, intratumoral immunoglobulin (Ig) A, an antibody predominantly expressed by isotype-switched B cells at mucosal locations, can increase anti-tumor immunity through coordinated T cell and B cell responses ( 9 ). Consistent with these mechanisms, several studies found that intratumoral B cells and tertiary lymphoid structures made up of B and T cells improve ovarian cancer survival ( 8 , 10 ). In an analysis that included data from the Nurses’ Health Study (NHS) and NHSII, greater tumor levels of antibody-producing plasma cells, a subtype of B cells, and IgA-coated cells were associated with improved ovarian cancer outcomes ( 9 ). The protective role of humoral responses in cancer has also been reported for other carcinomas and premalignant conditions, including endometrial and non-small cell lung cancer, as well as endometriosis ( 11 – 13 ). It is plausible that cigarette smoke influences ovarian cancer outcomes by impacting both the T cell and B cell anti-tumor response; however, data linking cigarette smoking with intratumoral B cell responses are lacking. Prior studies have shown that cigarette smoke exposure may lead to dysregulation and impairment of circulating B lymphocytes ( 14 ). Also, some B cells express nicotine receptors, and long-term exposure to nicotine can inhibit circulating B cell development and function ( 14 ). Smoke exposure may also alter abundance of IgA and IgG in serum and saliva, although results have been varied across populations ( 14 – 20 ). Further, in lung cancer patients, smoking was associated with higher plasma cell infiltration, particularly those cells that had IgG positivity ( 21 ), and more mature tertiary lymphoid structures ( 22 ), while pancreatic cancer patients who smoked had increased infiltration and diversity of IgG in their tumors ( 23 ), supporting that cigarette smoking can influence humoral immunity. Thus, we conducted a study to evaluate the hypothesis that cigarette smoke exposure in early life and adulthood was related to the ovarian tumor immune infiltration of B cells, including several B cell sub-populations, or Ig positivity. We also assessed risk of ovarian cancer for tumors with higher versus low B cell and Ig infiltration. Better understanding the link between cigarette smoke exposure and B cell responses may help identify tailored ovarian cancer treatment options (e.g., immunotherapies that target B cells) for patients with a history of smoking.