Tobacco Smoking and Survival Following a Diagnosis with Ovarian Cancer

Ovarian cancer is the most fatal gynecologic malignancy in the United States (1) with a 5-year survival rate of 48% (2). Consideration of whether lifestyle factors influence survival of patients with ovarian cancer may provide new opportunities for intervention. Tobacco smoke is a highly proinflammatory and toxic substance (3), and is a cause of premature death from multiple conditions, including cancer (4–6). In general, previous studies among women with ovarian cancer observed increased all-cause mortality among former and current smokers compared with never smokers (7–12). However, most of these studies, including our prior work (12), were unable to evaluate the influence of both active and passive smoking or adjust for residual disease following debulking surgery, an important independent prognostic factor (13). Moreover, few studies have examined the impact of smoking on ovarian cancer recurrence despite its potently tumorigenic effects and potential to interact with cancer treatments. To gain deeper insight into smoking as a modifiable factor that could potentially influence survival among women diagnosed with ovarian cancer, we investigated the associations of active and passive smoking with mortality in a hospital-based study.

In total, 862 patients with ovarian cancer visited Moffitt Cancer Center between 2008 and 2019. Eligible participants for this analysis included 519 cases (62%) who completed an electronic patient questionnaire with detailed information on demographics, lifestyle, medical and medication history, and cigarette smoking within six months of ovarian cancer diagnosis [median assessment time was the same month of diagnosis, interquartile range (IQR): 0 to 1 month after diagnosis; Fig. 1]. Tumor-related clinical variables (e.g., tumor stage, histology, cause of death) were obtained via medical chart abstraction or by linkage with cancer registry data. Neoadjuvant chemotherapy use, debulking status, chemotherapy regimens, platinum sensitivity, cause of death, ovarian cancer recurrence, and NSAID use were obtained from medical record abstraction for 181 patients who completed their entire treatment course at Moffitt Cancer Center.

The institutional review board (IRB) determined that the research project (MCC# 20389/Advarra IRB# Pro00040222) was exempt from oversight and met the criteria for and granted a Waiver of Health Insurance Portability and Accountability Act authorization.

Patients reported whether they had smoked at least 100 cigarettes during their entire life, current smoking status, and whether they had ever been frequently exposed to others’ cigarette smoke at home or at work. Both current and former smokers reported the age when they started smoking regularly, total years smoked, and the number of cigarettes smoked per day currently (current smokers) or before quitting (former smokers). Body height and weight were measured at the first visit to Moffitt Cancer Center and then were updated during treatment and follow-up.

The primary outcome was all-cause mortality among patients with ovarian cancer, while the secondary outcome was ovarian cancer–specific mortality. Deaths were identified via linkage to the National Death Index.

Cox proportional hazards regression models were used to estimate HRs and 95% confidence intervals (CI) for the association between smoking and survival. The analytic time scale was measured in months from ovarian cancer diagnosis to death or the last date of follow-up. Primary models were adjusted for age at diagnosis (continuous), tumor stage (I, II, III, IV, unknown), epithelial ovarian cancer histotype [high-grade serous/poorly differentiated, low-grade serous/non-serous (e.g., mucinous, endometrioid, clear cell, or borderline), unknown/other histology], and body mass index (BMI; <25, 25–29, ≥30 kg/m², unknown).We evaluated whether additionally adjusting for personal smoking status when evaluating exposure to secondhand smoke altered the risk estimates; however, this was not included in the final model as adjustment for it did not substantially change risk estimates., We also assessed whether there was a multiplicative interaction term between personal smoking and secondhand smoke exposure. In addition, we assessed the association of smoking with 5-year all-cause mortality.

We conducted a stratified analysis by age at diagnosis and multiple sensitivity analyses: (i) restricting to women with high-grade serous/poorly differentiated histology; (ii) restricting to stage III/IV cases; (iii) excluding mucinous ovarian cancer cases, for whom smoking is an established risk factor (14, 15); (iv) excluding women who died within the first month after diagnosis; (v) excluding women with missing tumor stage or stage IV, as these patients had much worse survival (16). We secondarily evaluated the impact of smoking with additional adjustment for debulking status (suboptimal, optimal), neoadjuvant chemotherapy (no, yes), or ever NSAID use (no, yes) among 181 women with these data. Within this subpopulation, we also assessed the associations of smoking with ovarian cancer–specific mortality and recurrence-free survival. All-cause mortality among those patients who experienced a recurrence during follow-up was also assessed, with post hoc sensitivity analyses restricting to platinum sensitive cases and adjusting for second-line chemotherapy regimens (platinum based vs. non-platinum based). All statistical tests were two sided and conducted with SAS statistical software version 9.4 (SAS Institute Inc.).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

In total, 519 ovarian cancer cases provided data on their smoking status and were included in the analyses, with a median age of 63 years (IQR: 55–70) and median survival time since ovarian cancer diagnosis of 2.8 years (IQR: 1.3–4.5; Supplementary Table S1). Most participants were diagnosed with high-grade serous or poorly differentiated carcinoma and stage III cancer; 85% were former or current smokers (Table 1). Compared with never smokers, current smokers were more likely to be diagnosed at a younger age with mucinous and/or early-stage (stage I/II) tumors, have a higher BMI, undergo treatment with neoadjuvant chemotherapy, and have optimal debulking surgery, but with a shorter overall survival time.

Compared with never smokers, the risk of all-cause mortality was increased by 70% among current smokers (HR = 1.70; 95% CI: 1.09–2.63), especially for those who smoked ≥15 cigarettes per day (HR = 1.92; 95% CI: 1.15–3.20; Table 2). This association was stronger among stage III/IV cases (current vs. never smokers, HR = 1.87; 95% CI: 1.12–3.12), but weaker for high-grade serous/poorly differentiated histology (HR = 1.39; 95% CI: 0.83–2.32; Table 3). Smoking duration ≥20 years (HR = 1.38; 95% CI: 0.94–2.03) and ≥20 pack-years (HR = 1.37; 95% CI: 0.91–2.05) were suggestively associated with greater risk of mortality compared with never smokers. This was stronger for stage III/IV cases [comparable HR and 95% CI: 1.53 (0.99–2.37) and 1.56 (0.99–2.45), respectively]. No association was observed for smoking initiation before age 18 (HR = 1.22; 95% CI: 0.80–1.85). When former smokers were excluded from analyses, smoking duration ≥20 years (HR = 1.74; 95% CI: 1.09–2.77), ≥20 pack-years (HR = 1.87; 95% CI: 1.12–3.10), and smoking initiation before age 18 (HR = 1.95; 95% CI: 1.13–3.37) were associated with increased risk of mortality compared with never smokers. Furthermore, neither environmental smoke exposure at home (HR = 1.16; 95% CI: 0.76–1.78) or at work (HR = 1.10; 95% CI: 0.75–1.60) was associated with risk of all-cause mortality overall or among those with high-grade serous carcinoma (at home HR = 1.24; 95% CI: 0.75–2.04; at work HR = 1.12; 95% CI: 0.72–1.73) or stage III/IV disease (at home HR = 1.26; 95% CI: 0.78–2.06; at work HR = 1.13; 95% CI: 0.72–1.76).

Compared with the overall study population, women with data on debulking status and usage of neoadjuvant chemotherapy who received their entire first-line treatment course at Moffitt Cancer Center were more likely to have longer survival time, higher BMI, and be ever smokers (Supplementary Table S1). Among these 181 women, the HR of current versus never smoking with all-cause mortality was 2.83 (95% CI: 1.00–7.98; Supplementary Table S2). Results were similar after adjustment for debulking status or neoadjuvant chemotherapy, but stronger when including NSAID use in the model (comparable HR = 3.34; 95% CI: 1.15–9.71). In this group, compared with never smokers, current smokers had a 2.73-fold higher risk of ovarian cancer–specific mortality (95% CI: 1.09–6.83; Supplementary Table S3), with similar results by number of cigarettes smoked per day.

Current smoking was not related to recurrence-free survival (current vs. never and past smokers, HR = 1.46; 95% CI: 0.86–2.48). Among patients who had a recurrence, smoking was related to a higher all-cause mortality (Fig. 2; comparable HR = 2.79; 95% CI: 1.44–5.41), with a similar though nonsignificant association among those without cancer recurrence (comparable HR = 2.78; 95% CI: 0.81–9.57). Results did not differ when restricting to patients who were platinum sensitive at the time of first-line or second-line treatment or with additional adjustment of second-line chemotherapy regimens.

Findings did not differ when excluding women who died within the first month after diagnosis (n = 2), mucinous tumors (n = 27), stage IV/missing stage tumor (n = 188) or when evaluating associations with 5-year all-cause mortality. In stratified analyses, we observed stronger associations among women diagnosed after versus before age 63 [e.g., current vs. never smokers, HR (95% CI) = 1.81 (0.92–3.56) and 1.27 (0.69–2.33), respectively].

The results of this retrospective, hospital-based study suggest that active smoking, but not secondhand smoke, is associated with worse ovarian cancer outcomes, even after adjusting for neoadjuvant chemotherapy and residual disease after debulking surgery. In addition, current smoking was also positively associated with risk of mortality among patients with ovarian cancer recurrence, despite the null association between smoking and risk of recurrence.

Utilizing data from two large prospective cohorts, the Nurses’ Health Study (NHS) and NHSII, our group previously observed increased mortality among patients with ovarian cancer who were current smokers (12). Results in the current study as well as several other prior studies were consistent with this finding (7, 8, 10, 17), including a large pooled analysis of 19 case–control studies in the Ovarian Cancer Association Consortium (11). Importantly, our study was one of the first to demonstrate that the association of current smoking with mortality remained even after adjusting for key predictors of survival, neoadjuvant chemotherapy, and debulking status. Furthermore, less is known of the impact of smoking on risk of recurrence among women with a diagnosis of ovarian cancer and how smoking interacts with treatment. Among the limited studies of smoking and ovarian cancer recurrence, inconsistent results have been reported. Smoking was not related to progression-free survival in a pooled analysis of 13 retrospective studies (18), which is similar to our findings with recurrence. Nevertheless, an inverse association of smoking with progression-free survival was observed in three smaller studies for ovarian cancer overall (19), clear cell carcinoma (20), or low-grade serous carcinoma (21). Overall, smoking is likely not strongly related to progression-free survival, although a modest association with specific subtypes cannot be ruled out. Unique to our study was the observation that current smoking was associated with worse survival among women who experienced a recurrence, even after adjusting for second-line therapy. This may suggest that smoking could influence the aggressiveness of a recurrent tumor or negatively interact with therapies reserved for the recurrent setting. Large prospective studies that have sufficient power to conduct both histology-specific analyses and interactions with different types of treatment are needed to further elucidate these relationships.

Several possible biologic mechanisms have been postulated regarding the relationship between smoking and cancer-related outcomes. For example, smoking may alter systemic and tumor physiology that can lead to complications associated with cancer treatment, continued development of comorbid disease, and activation of tumorigenic pathways (22). For patients with existing cancer, long-term smoking may result in a tumor milieu enriched with proinflammatory cytokines and chemokines, providing a preferred microenvironment for cancer progression (23). Smoking can also increase proliferation and tumorigenesis through the activation of cellular signaling pathways, drug-metabolizing enzymes, and tobacco-related genetic and epigenetic alterations (19, 22, 24–26, 27, 28). This may lead to a more aggressive behavior and a decreased response to chemotherapy, ultimately impacting survival for patients with ovarian cancer (19, 22, 24–26, 27, 28). Specifically, the numerous chemicals in cigarette smoke can interact with chemotherapeutic or other pharmacologic agents (29, 30), increasing the risk of treatment failure (31) and requiring adjustments to achieve an effective chemotherapy dose (32). Studies in other types of cancer showed decreased effectiveness of platinum-based chemotherapy in smokers, which is the most common first-line chemotherapy regimen in ovarian cancer (27, 28). In our study, there was no difference seen in platinum sensitivity between current and never smokers. However, we had a relatively small number of patients with data on platinum sensitivity, thus further studies with larger sample size are warranted. Moreover, smoking may impair glucose metabolism and cause myocardial infarction, which are all common among patients with ovarian cancer and adversely impact survival (33–36), although we observed similar associations for ovarian cancer–specific and all-cause mortality.

Strengths of this study include availability of detailed clinical data regarding the usage of neoadjuvant chemotherapy, surgical debulking status, cause of death, and cancer recurrence. The single-institutional nature of the study enabled uniform data collection, and more similar treatment practices among physicians, decreased intersubject variability, although this also may reduce generalizability. Limitations of this study include possible misclassification due to self-reported smoking, limited power to conduct detailed histology-specific analyses or stratification by residual disease after debulking surgery, and the inability to evaluate the impact of smoking on survival in response to second-line chemotherapy or bevacizumab, a mAb to VEGF (37–39). As Moffitt Cancer Center is a referral center with a large catchment base, many patients opt to have their surgery at Moffitt but receive their chemotherapy elsewhere to reduce travel burden. This led to exclusion of participants without complete cancer treatment–specific data in some analyses. Finally, most participants were in birth cohorts (median birth year 1948; IQR, 1941–1957) with a higher smoking prevalence, possibly limiting applicability of our findings to other patient populations with lower smoking rates (40, 41).

This hospital-based study provides evidence that smoking adversely impacts ovarian cancer survival, particularly after ovarian cancer recurrence, but not the probability of recurrence per se. Our findings support inclusion of smoking in prognostic prediction models as well as future studies of the effectiveness of adding structured smoking cessation programs in standard clinical management of patients with ovarian cancer, especially in recurrent settings. Further investigations in large population-based studies with detailed information on ovarian cancer treatments, recurrence, and platinum sensitivity are required to confirm these findings and further elucidate the underlying mechanisms by which smoking impacts ovarian cancer survival.

J.-Y. Chern reports personal fees from AstraZeneca, Seagen Inc, and Cigna/Carecore outside the submitted work. S.S. Tworoger reports grants from Moffitt Cancer Center during the conduct of the study; grants from NIH/NCI, State of Florida, DOD, and BMS; personal fees from AACR, Ovarian Cancer Research Alliance, Ponce Health Sciences University, German Cancer Research Center, and Harvard T.H. Chan School of Public Health outside the submitted work; and is a member of external advisory committee (non-paid relationships) of California Teachers Study (City of Hope) and The Tomorrow Project (Alberta Cancer Center). No disclosures were reported by the other authors.

T. Wang: Conceptualization, formal analysis, funding acquisition, writing–original draft, writing–review and editing. S.H. Read: Conceptualization, funding acquisition, writing–original draft, writing–review and editing. D. Moino: Data curation, writing–review and editing. Y. Ayoubi: Data curation, writing–review and editing. J.-Y. Chern: Conceptualization, resources, writing–review and editing. S.S. Tworoger: Conceptualization, resources, supervision, project administration, writing–review and editing.

This work was supported in part by the Junior Scientist Research Partnership Award at the H. Lee Moffitt Cancer Center and Research Institute.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.