Overall cancer risk
Evidence regarding cancer risk in PJS primarily comes from small cohorts and meta-analyses [15]. Some findings are consistent across studies, such as the high incidence of early-onset cancers with a median age at diagnosis below 50 years (48 years in our cohort, 42 years in the largest meta-analysis [15]). However, many of these studies are affected by recruitment bias, potentially leading to overestimated cancer risks [16]. Most cohorts are retrospective [17], and only a few include prospectively followed patients [5]. Furthermore, diagnoses are often based on clinical criteria, with only a subset of patients having confirmed STK11 mutations [3, 17].
In this study, we aimed to overcome these limitations by including only families with a pathogenic STK11 variant. For risk estimation, we used the GRL method, a statistical approach that accounts for unobserved genotypes (i.e. untested individuals) and corrects for ascertainment bias. These methodological differences likely explain the lower lifetime cancer risk observed in our cohort (18.1% in men and 36.8% in women, compared to 70–85% in previous studies [3, 7, 18]). Notably, when using the Kaplan-Meier method, which does not infer unobserved genotypes, the lifetime risk of cancer was 61% (95%CI: 34%–77%) for men and 66% (95%CI: 41%–80%) for women.
Mortality
Few studies have assessed life expectancy in PJS. In a cohort of 113 patients [19], increased mortality was reported (HR = 3.50), with 67% of deaths being cancer-related. Another study comparing PJS patients to age-matched controls found a threefold increase in mortality (HR = 3.4, 95% CI 1.5–7.9), whereas no excess mortality was observed for juvenile polyposis syndrome using the same methodology [20]. In our cohort, using SMRs, we confirmed that mortality is increased in PJS, particularly among women under 50 years of age. These findings suggest that cancer screening, especially for gynecological malignancies, could be improved.
PDAC
The incidence of PDAC in PJS remains controversial, with risk estimates at age 70 ranging from 11 to 36% and relative risks as high as 132 [3,4,5,6, 18]. A study by Korsse et al, dedicated to PDAC risk in PJS, reported a cumulative risk of 26% (95% CI 4–47%) at age 70 and a RR of 76 [5].
Due to these high estimates, most guidelines including those from the Cancer of the Pancreas Screening (CAPS) Consortium recommend pancreatic surveillance in PJS [7, 21, 22]. According to the CAPS guidelines, MRI/MRCP or endoscopic ultrasound should be used annually in PJS patients without pancreatic lesions.
In our study, we report a lower PDAC risk of 4.5% (95CI: 0.8%–10%) at age 70 and a RR of 5.5. Notably, one case initially classified as PDAC was reclassified as duodenal adenocarcinoma with pancreatic invasion upon reanalysis of the pathology report. Similar misclassifications have been reported [5, 23] and may contribute to overestimated PDAC risk in PJS.
The optimal age to begin pancreatic surveillance in PJS is debated. In the CAPS consortium, most experts recommend starting at age 40, although 32% suggest earlier initiation [21]. In our cohort, the median age at diagnosis of PDAC was 60 years, compared to 54 years in Korsse et al. [5]. However, early cases were observed in both cohorts (two before age 50 in ours, and three of seven in Korsse et al., including diagnoses at ages 35 and 36). Based on their observations, Korsse et al. recommend starting pancreatic screening in PJS patients at the age of 30 years. Given the median age at diagnosis of 60 years in our cohort, we suggest that starting pancreatic surveillance at 40 years old would be a reasonable compromise, as this would reduce the number of examinations while limiting the risk of overlooking early-onset cases.
Importantly, only one of our 8 PDAC patients had a family history of PDAC, consistent with Korsse et al., where none of the seven cases had a family history of PDAC [5].
We found that IPMN is highly prevalent in PJS, with a cumulative incidence of 65% at age 70. In the general population, the prevalence of IPMN lesions ≥2 mm on pancreatic MRI is around 49% in patients > 50 years of age [24], but lesions ≥10 mm are much less common than in our cohort (around 3%). Degenerated IPMN have been reported by other groups in PJS [5], and somatic STK11 loss is common in IPMN outside the context of PJS [25, 26]. The high proportion of PDAC cases originating from degenerated IPMNs in this cohort suggests that degenerated IPMNs may be an important contributor to PDAC risk in PJS. For comparison, this pathway accounts for 10% of PDACs overall [27]. Importantly, pancreatic surveillance may bias this correlation by improving the detection of small IPMNs. However, our findings contrast with those of a large cohort of patients at high risk of PDAC who also underwent pancreatic surveillance [28]: in this cohort, none of the 10 PDAC cases was associated with IPMN. The surveillance of patients with IPMN is guided by the CAPS and Kyoto (formerly Fukuoka) guidelines [14]. In patients at high risk for PDAC, a cyst growth rate >2,5 mm/year was shown to predict malignant transformation [29]. Indeed, in our cohort, all patients who developed PDAC had a growth rate >2,5 mm/year, suggesting that growth rate could inform surveillance intervals in PJS. However, these findings require confirmation in independent cohorts.
Taken together, these data support pancreatic surveillance in PJS, despite a lower PDAC risk than previously reported. We recommend yearly pancreatic screening from age 40, with particular attention to IPMN detection.
Other digestive tract cancers
PJS patients are at increased risk for colorectal cancer, although the relative risk in our cohort was lower than in previous studies [4, 18]. In addition to the methodological biases discussed above, this may be due to endoscopic polyp removal [7, 15]. Indeed, a Chinese retrospective study reported more colorectal cancer cases in rural areas than urban ones, attributing this to limited access to care [30]. In our cohort, only one in five colorectal cancer cases occurred in patients under endoscopic surveillance. Small bowel cancer, although extremely rare in the general population, was diagnosed in 10 PJS patients, with a relative risk of 19.0 at age 40. All but one case originated in the duodenum, underscoring the importance of thorough duodenal evaluation during endoscopic surveillance. Small bowel screening is also recommended in PJS to prevent mechanical obstruction from large polyps.
Non-gastrointestinal cancers
In this study, we confirm that women with PJS are at high risk for pelvic tumors. For women, the lifetime risk of gynecological cancer (ovary or cervix) exceeded that of digestive tract cancers. These malignancies were diagnosed at younger ages than gastrointestinal cancers (median age: 44.5 years for gynecological cancers, 55 years for breast cancer). Early-onset cervical and ovarian cancers contributed significantly to excess mortality in women under 50.
Consistent with previous reports, ovarian and cervical cancers were of rare histologic subtypes [31]. All cervical cancers were gastric-type adenocarcinomas (also known as minimal deviation adenocarcinomas), which are HPV-independent and have a poorer prognosis than typical cervical adenocarcinomas. These findings highlight the importance of gynecological surveillance in young women with PJS. Our current PRED-IdF guidelines recommend initiating pelvic screening at age 25, including annual clinical examination, pelvic ultrasound, and cervical smear every three years. It is crucial that patients and physicians are informed about this screening, and that gynaecologists managing PJS patients are aware that they should search for non-HPV-related lesions. We consider prophylactic gynecological surgery (radical hysterectomy) to be an acceptable option, and this should be discussed on a case-by-case basis. Given the early onset of these cancers reported here, it is probably reasonable to consider this surgery from the age of 40.
We also confirm an increased risk of breast cancer in women with PJS. However, the relative risk does not exceed 4 : according to our national guidelines [32], this is a high risk but not a very high risk (as in the hereditary breast and ovarian cancer (HBOC) syndrome, for example). The excess risk was even lower after age 50 (RR of 1.6 at age 70). These findings contrast with older studies reporting very high risks [4, 18]. Most breast surveillance guidelines for PJS mirror those for the HBOC syndrome. For example, the European Hereditary Tumor Group (EHTG) [7] and NCCN [33] guidelines recommend annual screening starting at age 25 and age 30, respectively, including clinical examination and mammography, or annual breast MRI with and without contrast for dense breasts. Further studies are needed to determine whether such intensive follow-up is warranted, as it seems excessive in light of our results.
In our cohort, the ratio of breast and gynecological cancers to gastrointestinal cancers was high (27 breast and gynecologic cancers vs. 17 gastrointestinal cancers), whereas the largest meta-analysis on PJS reported 99 and 198 cases, respectively [15]. This discrepancy may reflect reporting bias in earlier studies, as PJS patients are often followed by gastroenterologists, and families without gastrointestinal malignancies may be underrepresented.
Another notable finding is the increased risk of lung adenocarcinoma. Importantly, this finding should be confirmed in independent cohorts with better adjustment for smoking exposure. Although this risk was previously described in PJS [19], lung cancer screening is not routinely recommended in PJS. The American College of Gastroenterology guidelines suggest considering chest computed tomography (CT) in smokers [22]. The development of the low-dose CT scan may prompt us to reconsider lung cancer screening in the next update of our guidelines, particularly for smokers. At minimum, smoking cessation should be strongly encouraged.
Limitations
This study has several limitations. As PJS is a rare condition, the number of patients included is relatively small. Moreover, although the GRL method accounts for intrafamilial correlation by inferring unknown genotypes from the family structure, this correction inevitably reduces statistical power, given that only 37 families were informative for risk estimation. Nevertheless, this study relied on a genuine cohort with a median number of 13 individuals per family, representing 576 individuals and a substantial number of cancer events. Risk estimates were stable, as reflected by the bootstrap-derived 95% confidence intervals. Despite these strengths, the limited number of informative families may not fully capture the between-family variance of PJS, and therefore confirmation in larger cohorts will be essential. Importantly, a pathogenic variant in STK11 was identified in every family, whereas several other cohorts have included kindreds without confirmed STK11 mutations.
The mixed prospective–retrospective design of this study may introduce differential biases, including under‑ascertainment or misclassification of cancer events in retrospectively collected data, and survivorship bias for cancers with a poor prognosis. To mitigate these risks, retrospective data were collected by trained medical geneticists and most reported cancers underwent systematic verification from medical records (pathology reports, operative notes, and imaging summaries). For the prospective component, the regular follow‑up of PJS patients enabled centralized documentation, which supported a high degree of completeness. While these measures strengthen data quality and reduce ascertainment bias, we acknowledge that residual bias cannot be entirely eliminated.
The median age of patients in the prospective follow-up group is relatively low. This is due to the fact that our network was established in 2009 and did not include some of the patients who had previously undergone genetic testing and were followed in endoscopy centers outside the network. This reduces the contribution of the prospectively followed patients to the overall cancer risk estimates. An update of these results in the coming decade would help strengthen the findings.
Finally, relative risk calculations were based on cancer incidence and mortality estimates from the period 1990–2018. However, cancer incidence evolves over time due to environmental factors such as smoking habits and diet. For instance, the incidence of PDAC is rising in Europe. These temporal changes may influence cancer risks in PJS patients: for example, countries with low breast cancer incidence, such as China or Japan, also report lower breast cancer rates in PJS cohorts [3, 30]. A Japanese meta-analysis identified a time-dependent trend, with higher rates of gynecological malignancies in more recent cohorts compared to older studies [3]. However, data on environmental exposures were not available in our cohort.
