Pelvic inflammatory disease and risk of borderline ovarian tumors: A national population-based case-control study in Sweden.

Sarah Jonsson: Conceptualization; formal analysis; investigation; methodology; project administration; visualization; writing – original draft. Håkan Jonsson: Conceptualization; writing – review & editing. Eva Lundin: Funding acquisition; writing – review & editing. Christel Häggström: Formal analysis; writing – review & editing. Annika Idahl: Conceptualization; funding acquisition; methodology; supervision; writing – review & editing.

This study was approved by the Swedish Ethical Review Authority, Dnr 2020‐03679.

This investigation was supported by grants from the Lion's Cancer Research Foundation (LP23‐2343, LP 22‐2315, LP 21‐2280, LP 20‐2233, LP 18‐2196), Umeå University, and Insamlingsstiftelsen, the Medical Faculty, Umeå University. The funding sources were not involved in study design, collection, analysis, interpretation, writing, or the decision to submit the article for publication.

This is a national population‐based case–control study using Swedish nation‐wide register‐based data. The cases consist of all women residing in Sweden and diagnosed for the first time with BOT between January 1, 1999, when ICD‐10 was fully implemented in Sweden, and December 31, 2020. Cases were matched to 10 controls each at date of diagnosis (index date), on year of birth and residential district by the Statistics Sweden (SCB). Data on potential confounders were extracted from Swedish nationwide registers. With the help of the personal identification number which citizens in Sweden receive at birth or immigration, a unique linkage between different registers is possible. 22 Additional descriptions of the Swedish national registries included in the study can be found in the Supplementary Materials. Women with BOT were selected from the Swedish National Cancer Register with respect to tumor site using the International Classification of Diseases for Oncology, second revision (ICD‐O/2); ovary (C56.9), fallopian tube (C57.0), peritoneum (C48.1‐2), abdomen and pelvis (C76.2‐3) (Table  S1 ). We chose to include borderline tumors from the fallopian tube and the peritoneal cavity since they closely resemble their ovarian counterparts. 23 , 24 Cases with BOT classification was included. Women with tubo‐ovarian carcinoma at or before index date were excluded. Tumor histotypes were specified using the morphology codes of ICD‐O/2 (Table  S1 ). Information about tumor stage according to FIGO was retrieved for all cases from 2004 and onwards when tumor stage was introduced in the Swedish National Cancer Register. Control individuals were matched using the Swedish Register of the Total Population 25 and drawn without reversal; thus, a control individual could only occur once. Cases and controls with ovarian cancer or BOT before 1999 were excluded using the Swedish National Cancer Register. Women who had undergone the surgical procedures bilateral oophorectomy, bilateral salpingo‐oophorectomy or major gynecological surgery including bilateral salpingo‐oophorectomy before index date were also excluded using the National Patient Register (Table  S1 for detailed diagnostic codes). Furthermore, to increase the possibility that potential PID episodes were included in the National Patient Register the included women had to be resident in Sweden since the age of 18, or else excluded. If cases were excluded, the subsequent controls were excluded. We performed quality checks on data delivered from the National Board of Health and Welfare based on the above‐mentioned exclusion criteria. The selection process is described in Figure  2 . BOT diagnosed between 1999 and 2020. Information regarding clinically verified PID was retrieved from the National Patient Register. In the present study we defined PID as an inflammation of the female upper reproductive tract, including the diagnoses salpingitis, oophoritis and tubo‐ovarian abscess using the International Classification of Diseases version 9 (ICD‐9) and ICD‐10 (614A‐C and N70.0‐9, respectively). We also studied PID extended (PID ext ), including endometritis, salpingitis, oophoritis, pelvic peritonitis and tubo‐ovarian abscess defined by ICD‐9 and ICD‐10 (614A‐X, 615 and N70.0‐9, N71, N73‐4 respectively, Table  S1 ). PID was used as a never/ever variable and as number of PID episodes, categorized as (0, 1, 2, ≥3). To avoid reversed causation and misclassification, exposures less than 90 days prior to index date were not accounted for in the analyzes. Potential confounders were chosen a priori and included age, parity, and educational level (Supplementary Materials  S1 ). Parity at index date was extracted from the multi‐generation Register. Highest educational level, 1 year prior to index diagnosis year, was retrieved from the Longitudinal Integration Database for Health Insurance and Labor Market Studies. There is limited and contradictory evidence regarding previous gynecological surgery and risk of BOT. Previous gynecologic surgery could potentially be both a confounding factor and a mediator, therefore previous gynecologic surgery was not included in the main analysis. However, in an exploratory analysis it was accounted for. Previous gynecological surgical procedures such as salpingectomy, unilateral salpingo‐oophorectomy, hysterectomy, and tubal ligation were retrieved from the National Patient Register. Lateralization of salpingectomy was included for women having a surgical procedure before 1997. If the woman had a registered salpingectomy procedure twice during the whole study period, it was registered as bilateral salpingectomy. Gynecological surgical procedures within 90 days prior to index date were not accounted for in the analyzes. We categorized the covariates as number of children (0, 1, 2, ≥3); and educational level (mandatory school, high school, university graduate or other). Previous gynecological surgical procedures were categorized as follows; salpingectomy (unilateral/bilateral/unspecified) (yes/no), unilateral salpingo‐oophorectomy (yes/no), hysterectomy (yes/no), and tubal ligation (yes/no). For statistical analyzes IBM SPSS Statistics software version 26 was used. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated using conditional logistic regression on matching factors and adjusted for potential confounders (educational level and parity) (aOR). We performed analyzes stratified on the different histotypes. A test for trend was conducted for analysis of “number of PID” by treating it as a continuous variable. Previous gynecological surgery (salpingectomy, salpingo‐oophorectomy, hysterectomy, tubal ligation) in addition to parity and educational level were adjusted for in a separate analysis (aOR1). We performed sensitivity analyses where we included only borderline tumors of the ovary (C56.9).

The study population consisted of 4782 women with BOT and 45,167 controls (Figure  2 ). Table  1 presents demographic characteristics, exposure data, and confounding factors for both cases and controls. Supplementary Table S 2 provides additional background characteristics. Demographic, exposure, and tumor characteristics for cases with borderline ovarian tumor and matched controls. Abbreviations: BOT, borderline ovarian tumor; PID, pelvic inflammatory disease including salpingitis, oophoritis, and tubo‐ovarian abscess; PID ext , PID and additionally including endometritis and pelvic peritonitis. Numbers may not sum to total because of missing data. PID and surgical procedures within 90 days prior to index date were not accounted for in the analyzes. Case characteristics are presented in Table  1 . The average age of women diagnosed with BOT was 56.0 years (SD 16.0, range 18–95 years). Serous BOT emerged as the predominant histotype ( n  = 2953, 61.8%), followed by the mucinous histotype ( n  = 1739, 36.4%) (Table  1 ). In the present study, 94 of 4782 cases (2.0%) and 580 of 45,167 controls (1.3%) had one or more clinically verified PID (Table  1 ). Age at first PID did not differ between cases and controls (mean 37.2 vs. 36.7 years, respectively) (Table  1 ). Previous PID was associated with an increased risk of BOT overall (OR, 1.56; 95% CI, 1.25–1.94), also when adjusting for educational level and parity (aOR, 1.48; 95% CI, 1.19–1.85) (Table  2 ). Histotype‐specific analyzes showed an increased risk of serous BOT (aOR, 1.76; 95% CI, 1.36–2.29), but no association of PID with mucinous BOT (aOR, 0.95; 95% CI, 0.60–1.49) (Table  3 ). There was a significant trend toward increased risk of serous BOT with an increasing number of PID episodes (1, 2, ≥3; P trend  < .001) (Figure  3 ). Association between clinically verified pelvic inflammatory disease and risk of borderline ovarian tumor in women diagnosed with BOT year 1999–2020. Abbreviations: BOT, borderline ovarian tumor; PID, pelvic inflammatory disease including salpingitis, oophoritis, and tubo‐ovarian abscess. aOR, odds ratio, conditioned on matching factors (age, residential district) adjusted for educational level and parity. PID within 90 days prior to index date was not accounted for in the analyzes. Association between clinically verified pelvic inflammatory disease and borderline ovarian tumor in women diagnosed year 1999–2020 by histotype. Abbreviations: BOT, borderline ovarian tumor; PID, pelvic inflammatory disease including salpingitis, oophoritis, and tubo‐ovarian abscess; PID ext , PID and additionally including endometritis and pelvic peritonitis. PID within 90 days prior to index date was not accounted for in the analyzes. aOR, odds ratio, conditioned on matching factors (age, residential district) adjusted for educational level and parity. Number of pelvic inflammatory disease episodes categorized into 1, 2, 3 or more, odds ratio with 95% confidence interval conditioned on matching factors (age, residential district) adjusted for educational level and parity (aOR). PID, pelvic inflammatory disease; SBOT, serous borderline ovarian tumor. The results were similar when using the extended definition of PID (PID ext ), including endometritis, salpingitis, oophoritis, pelvic peritonitis, and tubo‐ovarian abscess; PID extended was associated with an increased risk of serous BOT (aOR, 1.66; 95% CI, 1.35–2.03) but not with mucinous BOT (aOR, 0.90; 95% CI, 0.64–1.26) (Table  3 ). Adjusting for previous gynecological surgery in addition to parity and educational level, did not alter the results; BOT overall (aOR1, 1.42; 95% CI, 1.13–1.80), serous BOT (aOR1, 1.61; 95% CI, 1.22–2.12) and mucinous BOT (aOR1, 0.97; 95% CI, 0.61–1.54). In a sensitivity analyses including only borderline tumors of the ovary (C56.9) the results where similar (Table S3 ).

In this nationwide register‐based case–control study, a history of PID was associated with an increased risk of BOT. Histotype‐specific analyzes revealed an elevated risk of serous BOT, but not for the mucinous histotype. Additionally, we identified a trend indicating an increased BOT risk with a higher number of PID episodes. These associations were robust and persisted regardless of whether PID was defined as salpingitis, oophoritis, and tubo‐ovarian abscess, or more broadly, with an extended definition of PID including endometritis and pelvic peritonitis. Only a few studies have previously investigated the association of PID with BOT. This is the largest study to date using clinically verified PID episodes and BOT diagnoses enabling the comprehensive histotype‐specific analyses. Furthermore, the effect on risk of repeated PID episodes could be carried out satisfactorily per histotype. A Danish retrospective population‐based case–control study utilizing self‐reported PID data found an overall association between PID and BOT However, no histotype‐specific associations were revealed. 18 Another study from the Ovarian Cancer Association Consortium, with pooled data on BOT from 11 studies, also used self‐reported, retrospectively collected PID data. That study found an increased risk of BOT, with higher odds ratio for BOT overall with ≥2 PID episodes. 20 There are few previous register‐based studies using clinically verified PID to investigate the association between PID and BOT. 21 , 26 Although no crude hazard ratios were presented, an increased overall BOT risk was observed when adjusting for parity as well as additionally adjusting for hysterectomy, tubal ligation, and endometriosis (data were not presented). 21 Concerning the association between PID and different histotypes, similar to Rasmussen et al., we observed an increased risk of serous BOT, but not mucinous BOT. 21 The variation in risk may be related to differing pathogenesis of the two tumor types. Serous BOT and serous carcinomas are proposed to originate in the fallopian tube, 6 , 27 while less is known about the origin of mucinous BOT. Another study has shown that a high frequency of chronic salpingitis is associated with serous BOT. 28 In our study, there was a significant trend towards an increased risk of serous BOT with an increasing number of PID episodes, corroborating the results of Rasmussen et al., 21 as well as a non‐significant association in the study from the Ovarian Cancer Association Consortium. 20 Despite the infrequency of PID diagnoses among women, this condition remains linked to an elevated risk of BOT, particularly, the serous histotype. Women with multiple episodes of PID appear to be at the greatest risk for subsequent serous BOT. While the association between prior PID and BOT risk is acknowledged, significant avenues for exploration remain. Establishing causality necessitates further investigation into the temporal relationship and underlying biological mechanisms. Although PID precedes BOT, the lead time between inflammation and tumor development is undetermined. It remains unclear whether the increased risk is due to the tumorigenic effect of a specific pathogen, the inflammatory process itself, or both. Notably, Chlamydia trachomatis , the major cause of PID, may have tumorigenic effects. 29 , 30 , 31 , 32 It is likely that additional, yet unknown, contributing factors exist and require further investigation. Additionally, the mechanisms of inflammation and tumor development should be studied more closely in various experimental models. Moreover, studies exploring whether accurate and timely treatment of PID can mitigate the risk for BOT should be conducted. Serous BOT and serous tubo‐ovarian carcinomas share similar reproductive risk factors and potential fallopian tube origin, suggesting overlapping etiological pathways. 1 , 2 , 4 , 6 However, despite these similarities, their clinical behaviors diverge, highlighting the need for further research to elucidate the mechanisms driving tumor differentiation to either tumor type, or progression from borderline to invasive forms. 3 , 5 , 9 The major strength of the present study is the study size; it is so far the most extensive study with the highest number of cases conducted on PID and the risk of BOT. Another strength is that the data are based on clinically verified register‐based exposures. We collected exposure information from the continuously updated high‐quality Swedish registries, thus eliminating recall bias compared to previous studies. 18 , 20 Additionally, these registries are nationwide, reducing the risk of selection bias. The registers include consistently recorded details on exposure, outcome, and covariates. Furthermore, we conducted analyses according to histotype, number of registered PID episodes, and controlled for potential confounders. Tumor morphology was available for the whole study population defined by ICD‐O/2 criteria. The exposure data encompassed diagnoses and surgical procedures from inpatient care from 1987 and specialized outpatient clinics since 2001. Diagnoses from general practitioners in outpatient settings were not available. Additionally, PID is challenging to diagnose, and many women may not seek medical care for it. 33 Consequently, the number of women diagnosed with PID might be underestimated, leading to potential misclassification. However, this would likely result in an underestimation of the association and the clinical impact of PID, as the misclassification is considered non‐differential. The Swedish Prescribed Drug Register, established in 2005, limited the ability to adjust for combined oral contraceptives and menopausal hormonal therapy across the entire study population. In this large nationwide register‐based case–control study encompassing clinically verified PID episodes, a history of PID is linked to an increased risk of serous BOT but not mucinous BOT. Additionally, a clear dose–response relationship is apparent. Future research should prioritize understanding the underlying biological mechanisms and evaluating the effects of PID detection and treatment to fully grasp this association.

The pathogenesis of pelvic inflammatory disease (PID) has garnered attention as a modifiable factor that may elevate the risk of epithelial tubo‐ovarian tumors, although current evidence remains constrained. Within the tubo‐ovarian tissues, different types of tumors can manifest ranging from benign and borderline to malignant. 1 While these conditions arise in the same organ, they represent distinct entities to some extent. 2 Borderline ovarian tumors (BOTs) exhibit disparities from malignant tumors in terms of histological architecture, growth pattern and prognosis. 3 Nevertheless, despite these distinctions there are notable parallels. BOT evolve from cystadenoma with stepwise mutations with potential to develop into low‐grade serous carcinoma (LGSC). Both serous BOT and serous invasive epithelial ovarian cancer (EOC) share similar reproductive risk factors, such as parity, breastfeeding, oral contraceptive use, and the timing of menopause. Although the strength of these associations may vary slightly between the two tumor types, these factors appear to influence the risk of both conditions. 4 , 5 Research suggest that serous BOT and serous carcinomas may have similarities in terms of origin and risk factors. Molecular and gene profiling studies indicate that, like the majority of EOC, both BOT and low‐grade serous carcinoma (LGSC) may originate from the fallopian tube epithelium. 1 , 2 , 3 , 6 , 7 , 8 , 9 This hypothesis is supported by recent findings demonstrating the tubal origin of these tumors, which aligns with observations of their shared risk factors. 7 Ascending microorganisms from the lower genital tract can cause inflammation of the upper reproductive organs, described as PID. It involves a spectrum of conditions including endometritis, salpingitis, oophoritis, pelvic peritonitis, and tubo‐ovarian abscess. Inflammation in the female upper reproductive tract has been suggested to be implicated in tubo‐ovarian tumorigenesis (Figure  1 ). 10 Chronic infection and inflammation are increasingly recognized as components of the oncogenic cascade, with microorganisms proposed to contribute to 15%–20% of all cancers. 11 Studies of previous PID as a risk factor for epithelial ovarian cancer have shown ambiguous results, some studies suggest an association 12 , 13 , 14 , 15 , 16 , 17 while others do not. 18 , 19 , 20 There are similar results with BOT, however the evidence is limited. 17 , 18 , 20 , 21 Recall bias for PID exposure is a concern in many of the studies, why the results must be interpreted with caution. Pelvic inflammatory disease salpingitis/oophoritis/tubo‐ovarian abscess. In parallel with previous studies investigating the association of PID with EOC, 12 , 13 , 14 , 15 , 16 this study aims to explore whether an association could be observed between clinically verified and register‐based PID and risk of BOT can be demonstrated in a population‐based case–control study using national Swedish registers.

The authors report no conflict of interest.

Data S1. Supporting Information.