Lifetime Exposure to Cigarette Smoke and Risk of Ovarian Cancer by T-cell Tumor Immune Infiltration

Cigarette smoke can lead to chronic inflammation and alter immune function, both locally in the lungs and bronchioles, as well as systemically (1–4). For example, exposure to cigarette smoke, either firsthand or secondhand, can alter the number of circulating CD4⁺ and CD8⁺ T cells (2, 5), leading to an increase in acute infections and chronic diseases such as chronic obstructive pulmonary disease and cancer (1, 3, 5, 6).

Specifically, for ovarian carcinogenesis, a personal history of smoking cigarettes has been associated with an increased risk of mucinous ovarian cancer for both former and current smokers, as well as an increased risk of highly aggressive ovarian cancer (death within 1 year of diagnosis), regardless of histotype (7–9). Moreover, we observed that women exposed to secondhand cigarette smoke in childhood and adolescence had a modest increased risk of ovarian cancer later in life (10). Furthermore, several studies found higher mortality rates following ovarian cancer diagnosis for former and current smokers compared with never smokers, with worse outcomes among those with high-grade serous tumors, suggesting that smoking may lead to more aggressive tumors (11–13).

Increasing evidence suggests that exposure to cigarette smoke can alter the tumor immune response for several cancers, including lung and colorectal tumors (14–19). High tumor-infiltrating T cells have been associated with improved prognosis of patients with ovarian cancer, particularly for high-grade serous carcinomas (20, 21); however, no studies have examined the relation of exposure to cigarette smoke in early life and adulthood with the tumor immune microenvironment of ovarian tumors. Therefore, we conducted a study among women in the Nurses’ Health Study (NHS) and NHSII evaluating the relationship of exposure to cigarette smoke during childhood and adulthood with risk of ovarian cancer based on the level of T-cell immune infiltrations, considering several T-cell subpopulations and immune checkpoints.

We leveraged two prospective cohorts, the NHS and NHSII. The NHS began in 1976 and included 121,700 U.S. female registered nurses age 30 to 55 at enrollment (22). The NHSII enrolled 116,429 female registered nurses age 25–42 in 1989 (23). Both cohorts 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 Institutional Review Board (IRB #00000971) and those of participating registries as required.

In 1982, NHS participants were asked: “Did your parents smoke while you were living with them?” and in 1999, participants in NHSII were asked: “When you were a child, did either of your parents smoke regularly inside your home?”. Response options to both questions were no, mother only, father only, or both mother and father. If the participant selected mother, father, or both, they were considered to have early life smoke exposure. To evaluate adult cigarette smoking behavior, participants in both cohorts were asked at baseline about their smoking history and in each subsequent questionnaire about their current smoking status. These responses were used to categorize the participant as a never, former, or current smoker at each questionnaire. Finally, to evaluate lifetime exposure to cigarette smoke, a composite variable was created using early life exposure and adult smoking [never smokers with no early life exposure, never smokers with early life exposure, ever smokers (including former and current) without early life exposure, and ever smokers with early life exposure].

Participants self-reported ovarian cancer diagnoses via questionnaire or were identified via linkage to the National Death Index (24, 25). Cases were confirmed by medical record review that included pathology reports or linkage with state cancer registries. A study pathologist reviewed pathology reports to obtain stage.

Formalin-fixed and paraffin-embedded ovarian tumor tissue samples were collected when information on the surgery was available. The most common reasons tumors were not collected were: tissue destruction by hospital prior to request, permission to request tissue could not be obtained because the participant was deceased, or hospitals were unable to send a sample (26). We obtained tumor tissue from 530 ovarian cancer cases, which were reviewed by a gynecologic pathologist to obtain grade, morphology, and histology and circle areas of tumor. These were used to create seven tissue microarrays (TMA) with three 0.6 mm cores per case that were immunostained using the AKOAYA Biosciences OPAL 7-Color Automation IHC kit. Two multiplex panels were analyzed; the first included T-cell markers CD3, CD4, CD8, FOXP3, and CD69, and the second included CD3 and four immune checkpoints including PD-1, PD-L1, Lag-3, and Tim-3. Both panels included pan-cytokeratins, a masking marker for tumor cells, and DAPI, which stains cell nuclei. All slides were imaged with the Vectra3 Automated Quantitative Pathology Imaging System. These images were spectrally unmixed via InForm (AKOYA) and exported to HALO (Indica Labs) for quantitative digital image analysis. A positivity threshold was determined for each marker using published staining patterns and intensity. The image was then analyzed for positive cells with staining in the cytoplasm or nucleus as relevant. We only considered immune cells that were infiltrating into areas of tumor. We excluded 22 cases from the first panel and 50 from the second panel without staining results due to damage or folding of the cores.

On the basis of prior literature of the relationship between smoking and tumor T-cell infiltration in other cancers (14–19, 27–29), our primary analysis evaluated early life and adult smoking with percentage (based on total tumor cells in each core) of total T cells (CD3⁺), Th cells (CD3⁺CD4⁺), cytotoxic T cells (CD3⁺CD8⁺), and regulatory T cells (CD3⁺CD4⁺FOXP3⁺). We then conducted an exploratory analysis evaluating T cells expressing the programmed death 1 protein (CD3⁺PD-1⁺), program death ligand 1 (PD-L1⁺), markers of T-cell exhaustion (CD3⁺Tim-3⁺ and CD3⁺Lag-3⁺), recently activated Th cells (CD3⁺CD4⁺CD69⁺), and recently activated cytotoxic T cells (CD3⁺CD8⁺CD69⁺) in relation to early life and adult smoking exposure. We used Tim-3⁺ and Lag-3⁺ as markers of T-cell exhaustion based on their importance biologically in suppressing immune response in ovarian cancer and target of immunotherapies as well as high performance on the immunofluorescence panels (30–32).

We excluded women with a bilateral oophorectomy, missing date of birth, menopause due to radiation, did not answer the question on early life smoking exposure, or had already died at baseline (1982 for NHS and 1999 for NHSII), and additionally excluded cases not included on the TMAs or were missing immune cell data. We included all epithelial ovarian cancer subtype and used competing risks Cox proportional hazards models to evaluate HRs and 95% confidence intervals (CI) for each variable with risk of ovarian cancer by each immune cell type listed above, dichotomized at the marker-specific median based on average core percent positive. Marker-specific medians were utilized as no clinical cut-off point has been set for immune cells; therefore, we used medians to maximize power. Models were adjusted for age, calendar year, cohort (NHS, NHSII) and known ovarian cancer risk factors, including menopausal status (pre, post), hormone therapy use (ever, never), oral contraceptive use (never, <1, 1–5, 5–10, 10+ years), parity (nulliparous, 1, 2, 3, >3), hysterectomy (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, and missing), updated by self-reported questionnaire every 2 to 4 years. Heterogeneity between the HRs for ovarian cancer with immune cell percentage below the median and the HR for ovarian cancer with immune cell percentage greater than or equal to the median was evaluated using the likelihood ratio test (33). In addition, we repeated analyses restricted to type II ovarian tumors (high-grade serous, poorly differentiated or transitional/Brenner). We were not able to separately examine other histotypes due to the limited number of cases.

To take advantage of the continuous nature the measured markers for the case-only analyses, we used beta-binomial models to estimate ORs and 95% CIs, interpreted as the ratio of odds that a tumor cell was positive for the marker(s) of interest in the exposed versus unexposed groups. Beta-binomial models are recommended in the setting of outcome measures with a zero-inflated distribution, as is the case with some immune cell types assessed in this study (34). Three to six cores were included on the TMA per tumor, so participant was included as a random effect in the model (34), and models were adjusted for fixed effects of: CD3⁺ tertiles (except in the analysis of CD3⁺ cells), age at cancer diagnosis, year of cancer diagnosis, histotype [type I (low-grade serous, endometrioid, clear cell or mucinous) versus type II (high-grade serous, poorly differentiated or transitional/Brenner)], and the covariates listed above. We conducted secondary analyses among type II tumors only, though due to the smaller sample, we could not adjust for parity (>89% of women in NHS were parous). Given the multiple immune cells being evaluated, we calculated the number of effective independent tests to be 6, resulting in a corrected P value of 0.0083 (0.05/6) to be considered statistically significant (35, 36). Analyses were conducted in SAS version 9.4 (SAS Institute) and R version 4.1.0 (R Core Team).

The data generated in this study are available upon request from the corresponding author.

We included 165,760 women in the NHS and NHSII who reported on early life exposure to cigarette smoke; 66% reported at least one parent smoked in the home (Table 1). During the follow-up period, 1,448 incident ovarian cancer cases were identified. Tissue was obtained and cores included in the TMAs for 530 cases (n = 1,590 tissue cores); image analysis was successfully completed for 508 cases on the T-cell panel and 480 cases on the immune checkpoint panel and complete early life smoke exposure data were available for 380 cases on the T-cell panel and 362 cases on the immune checkpoint panel. In line with previously published data in this cohort, cases included in the analysis were similar to all ovarian cancer cases identified, although cases in this study were slightly more likely to be diagnosed at an early stage due to the availability of tissue (Supplementary Table S1; ref. 37). Baseline ovarian cancer risk factors generally were similar for women by early life exposure to cigarette smoke, with women having early life exposure to cigarette smoke being slightly more likely to use oral contraceptives and more likely to ever have smoked compared to women without early life exposure.

Women who were exposed versus not to early life cigarette smoke had a higher risk of developing ovarian cancer with low levels of CD3⁺ T cells (HR: 1.50, 95% CI: 1.05–2.15, P = 0.03) while no association with risk of ovarian cancer with high CD3⁺ T cells was observed (HR: 1.01, 95% CI: 0.76–1.34, P = 0.96, P_heterogenity= 0.08; Fig. 1; Supplementary Table S2); however, this was not statistically significant at the Bonferroni-corrected α = 0.0083. The comparable risk estimates when restricting to type II tumors were 1.86 (95% CI: 1.12–3.11, P = 0.02) for low levels of CD3⁺ T cells and 1.00 (95% CI: 0.72–1.38, P > 0.99, P_heterogenity= 0.04) for high levels of CD3⁺ T cells (Supplementary Table S3), the association for low levels of CD3⁺ T cells was not statistically significant based on the Bonferroni correction. We also observed a similar association for cytotoxic T cells (CD3⁺CD8⁺) such that exposure to smoke early in life was suggestively associated with increased risk of developing tumors with low CD3⁺CD8⁺ cells (HR: 1.32, 95% CI: 0.96–1.82) overall and for type II tumors (HR: 1.47, 95% CI: 0.97–2.23, P = 0.07). However, there was no statistically significant heterogeneity in association levels of CD3⁺CD8⁺ T cells (P_{heterogeneity} > 0.2). There was no evidence of heterogeneity in associations of adult smoking status and lifetime exposure to cigarette smoke with risk of developing ovarian cancer by tumor infiltration of total T cells (CD3⁺), Th cells (CD3⁺CD4⁺), cytotoxic T cells (CD3⁺CD8⁺), or regulatory T cells (CD3⁺CD4⁺FOXP3⁺) among all tumors or type II tumors (Table 2; Supplementary Fig. S1; Supplementary Table S3).

When examining associations with continuous T-cell counts among cases only, there were no significant associations between exposure to early life cigarette smoke or adult smoking and odds of a tumor cell being positive for any of the markers of interest for all tumors or among type II tumors only (Table 3; Supplementary Table S4).

We assessed exhausted T cells (CD3⁺PD-1⁺, CD3⁺Tim-3⁺, CD3⁺Lag-3), recently activated helper and cytotoxic T cells (CD3⁺CD4⁺CD69⁺, CD3⁺CD8⁺CD69⁺), and overall PD-L1 (an immune checkpoint) in ovarian tumors in relation to smoking exposure in an exploratory analysis. We observed that exposure to smoke early in life was associated with increased risk of developing tumors with a low percentage of recently activated cytotoxic T cells (CD3⁺CD8⁺CD69⁺: HR: 1.45, 95% CI: 1.05–2.00, P = 0.03; Supplementary Table S5), but not a high percentage (CD3⁺CD8⁺CD69⁺: HR: 0.96, 95% CI: 0.71–1.31, P = 0.82). This association was stronger among type II tumors only (low CD3⁺CD8⁺CD69⁺ HR: 1.82, 95% CI: 1.16–2.85, P = 0.01; high CD3⁺CD8⁺CD69⁺ HR: 0.93, 95% CI: 0.66–1.31, P = 0.66; P_heterogenity= 0.02; Supplementary Table S6). In addition, among type II tumors, exposure to smoke early in life (HR: 3.19, 95% CI: 1.54–6.58, P = 0.002), ever smoking (HR: 2.48, 95% CI: 1.09–5.64, P = 0.03), or both (HR: 2.77, 95% CI: 1.35–5.67, P = 0.01) compared with never smoking and no early life exposure was associated with increased risk of developing tumors with low recently activated cytotoxic T cells (CD3⁺CD8⁺CD69⁺).

Among cases only, we observed that former smokers compared with never smokers had lower odds of having recently activated cytotoxic T cells (CD3⁺CD8⁺CD69⁺: OR: 0.71, 95% CI: 0.54–0.94), and ever smokers with early life exposure to smoke had significantly lower odds of having tumors with recently activated cytotoxic T cells (CD3⁺CD8⁺CD69⁺: OR: 0.64, 95% CI: 0.44–0.93; Supplementary Table S7). Similar findings were observed among type II tumors only (Supplementary Table S8). No other immune cell type or PD-L1 was associated with the smoking exposures.

In the first study to examine early life and adult smoking with the ovarian tumor immune microenvironment, we observed that exposure to cigarette smoke early in life was associated with an increased risk of specifically developing ovarian cancer with low levels of total T cells (CD3⁺) and a suggestive increased risk of developing ovarian cancer with low levels of cytotoxic T cells (CD3⁺CD8⁺), particularly among cases with type II tumors. This apparent association with cytotoxic T-cell infiltration appeared to be driven by a higher risk of ovarian cancer with low recently activated cytotoxic T cells (CD3⁺CD8⁺CD69⁺) for those with early life smoke exposure versus not. Interestingly, the associations with recently activated cytotoxic T cells were strongest for those with early life exposure to cigarette smoke and adult smoking compared with neither.

The associations we observed between early life cigarette smoke exposure and risk by total T cell and recently activated cytotoxic T cells did not reach statistical significance after Bonferroni correction, but overall associations for type II tumors, which are primarily comprised of high-grade serous and poorly differentiated carcinomas, were stronger. This may be due to enhanced immunogenicity of high-grade serous tumors, which are more likely to have T-cell infiltration compared with other histotypes (38). This finding is consistent with prior work from our group showing that smoking prior to diagnosis is mostly strongly associated with increased risk of death for those with high-grade serous tumors (13). It is possible that the increased DNA damage in high-grade serous carcinomas may act synergistically with the immune dampening effects of cigarette smoke (39, 40). Furthermore, one study has shown that exposure to smoke early in life may dampen the immune response through decreased percentage of CD8⁺ cells in the adenoids; however, more research is needed to confirm the impact of early life smoke exposure to abundance of total and cytotoxic T cells in the ovarian microenvironment (41). Furthermore, animal models of prenatal smoke exposure demonstrated that exposed male mice had a lower cytotoxic T-cell response when injected with lymphoma cells (42, 43). Additional research leveraging animal models, particularly for ovarian cancer, may elucidate key mechanisms and potential susceptibility to smoke exposure at an early age.

Despite observing associations with early life exposure to cigarette smoke, we did not observe that adult smoking behavior was clearly related to tumor immune infiltration. One possibility for this finding is the increased susceptibility of the immune system while it is still developing (44). In clinical models, secondhand smoke exposure was associated with suppression of T cell–dependent antibody response and suggested cigarette smoke exposure may interfere with the development of the immune system (45). However, studies in other adult cancers have observed associations between adult smoking and tumor T-cell infiltration. For example, a study among colorectal cancer tumors found that current and former smokers versus never smokers had an increased risk of colorectal cancer with low CD3⁺ cell density, and among non–small cell lung cancers, esophageal cancers, and head and neck squamous cell cancers, smokers had lower levels of intratumoral CD8⁺ T cells than never smokers (14–16, 27). Notably, smoking is a strong risk factor for these cancers, but is more weakly related to ovarian cancer (7, 10, 46–49). Furthermore, lung, esophageal, head and neck, and colorectal tumors are biologically different from ovarian cancer. For instance, non–small cell lung tumors are believed to be derived from stem cells in the bronchioalveolar duct junction, whereas ovarian cancers form most often from surface epithelial cells in the ovaries and fallopian tube (50, 51). To our knowledge, no studies have evaluated early life smoking exposure with tumor immune infiltration. Given the susceptibility of tissues to smoking during this developmental period, further research is needed to understand whether cigarette smoke exposure in childhood and adolescence impacts the tumor immune response for adult cancers (52).

This study has several strengths, including the large number of cases and extensive epidemiologic data. In addition, we used multiplex immunofluorescence to measure the immune markers, which allows measurement of up to six markers (plus DAPI) on one panel and assessment of colocalization of markers at the cell level, providing detailed information about T-cell subpopulations. However, this study has some limitations. First, we were not able to capture all facets of the tumor immune microenvironment. For example, tumor-associated macrophages have been associated with smoking status as well as ovarian cancer development and progression (53). Thus, further studies assessing a broader array of immune cell types are needed to fully understand the impact of smoking exposure on the ovarian tumor immune response (53–55). In addition, although the number of cases was relatively large, some of the T-cell subtypes were rare (e.g., CD3⁺CD4⁺FOXP3⁺, CD3⁺CD4⁺CD69⁺), thereby limiting our power in some analyses. Our measurement of early life cigarette smoke exposure was limited in that participants had to recall their exposure; however, we only analyzed participants who were diagnosed with ovarian cancer after reporting on early life exposure. Early life cigarette smoke exposure was asked dichotomously; therefore, we were unable to look at timing and duration of early life exposure. In addition, there may be measurement error in reports of early life smoke exposure; however, a prior study in the NHSII found good concordance between the participant's report of their mother smoking prenatally, and the participant's mother's self-report (56). Finally, our study was conducted among a predominantly white population of U.S. registered nurses, which limits generalizability of these findings to other racial, ethnic, and socioeconomic groups.

In summary, we observed that women exposed to cigarette smoke early in life had a moderate increased risk of developing tumors with low overall T-cell levels or low recently activated cytotoxic T cell levels while no association was observed with development of tumors with high levels of either cell type. These findings became stronger when specifically examining risk of type II tumors. This suggests that early life exposures may play a role in the immune response to tumorigenesis, although the underlying mechanisms are not well understood. Interestingly, few associations were observed with adult smoking and ovarian tumor T-cell infiltration, suggesting that smoking may adversely influence survival among patients with ovarian cancer through other immune pathways (e.g., macrophages) or through nonimmune mechanisms. Future research should evaluate exposure to cigarette smoke and ovarian cancer with other areas of the immune environment such as markers of inflammation and tumor associated macrophages. Finally, more work is needed to understand how timing of exposures throughout the life course may differently impact the tumor immune microenvironment.

C. Vinci reports grants from NIH outside the submitted work. J.R. Conejo-Garcia reports grants and personal fees from Anixa Biosciences; personal fees from Alloy Therapeutics and Mayflower Bioventures outside the submitted work; in addition, J.R. Conejo-Garcia has a patent for Anixa Biosciences issued, licensed, and with royalties paid from The Wistar Institute and a patent for Compass Therapeutics pending to Compass/Wistar. B.L. Fridley reports grants from State of Florida during the conduct of the study. S.S. Tworoger reports grants from Florida Department of Health, Department of Defense, and NIH/NCI during the conduct of the study; grants from Department of Defense, NIH/NCI, BMS; personal fees from AACR, Ponce Health Sciences University, Ovarian Cancer Research Alliance, NIH, and Roswell Park Comprehensive Cancer Center outside the submitted work; and non-paid relationships: member of external advisory committee of the UNC Lineberger Comprehensive Cancer Center, California Teachers Study (City of Hope) and The Tomorrow Project (Alberta Cancer Center). No disclosures were reported by the other authors.

C.A. Hathaway: Data curation, software, formal analysis, validation, investigation, visualization, writing–original draft. T. Wang: Data curation, software, formal analysis, methodology, writing–review and editing. M.K. Townsend: Data curation, software, supervision, visualization, writing–review and editing. C. Vinci: Conceptualization, funding acquisition, writing–review and editing. D.E. Jake-Schoffman: Funding acquisition, writing–review and editing. D. Saeed-Vafa: Resources, data curation, investigation, writing–review and editing. C. Moran Segura: Resources, data curation, investigation, writing–review and editing. J.V. Nguyen: Resources, data curation, investigation, writing–review and editing. J.R. Conejo-Garcia: Conceptualization, writing–review and editing. B.L. Fridley: Software, methodology, writing–review and editing. S.S. Tworoger: Conceptualization, resources, supervision, funding acquisition, visualization, methodology, project administration, writing–review and editing.

The authors would like to acknowledge the contribution to this study from central cancer registries supported through the Centers for Disease Control and Prevention's National Program of Cancer Registries (NPCR) and/or the NCI's Surveillance, Epidemiology, and End Results (SEER) Program. Central registries may also be supported by state agencies, universities, and cancer centers. Participating central cancer registries include the following: Alabama, Alaska, Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Florida, Georgia, Hawaii, Idaho, Indiana, Iowa, Kentucky, Louisiana, Massachusetts, Maine, Maryland, Michigan, Mississippi, Montana, Nebraska, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Puerto Rico, Rhode Island, Seattle SEER Registry, South Carolina, Tennessee, Texas, Utah, Virginia, West Virginia, Wyoming. The authors would also like to acknowledge the Channing Division of Network Medicine, Department of Medicine, Brigham and Women's Hospital as home of the Nurses’ Health Studies. The authors assume full responsibility for analyses and interpretation of these data. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The authors would like to acknowledge the Advanced Analytical and Digital Laboratory at Moffitt Cancer Center for their contribution in the multiplex immunofluorescence quantitative digital image analysis. T. Wang, M.K. Townsend, C. Vinci, D.E., Jake-Schoffman, J.R. Conejo-Garcia, B.L. Fridley, and S.S. Tworoger have been awarded the Florida Department of Health James & Esther King Biomedical Research Program grant (9JK02). We would also like to acknowledge the following grants: UM1 CA186107, P01 CA87969 (NHS), and U01 CA176726 (NHSII).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.