Article Text
Abstract
Background Peutz–Jeghers syndrome (PJS) is associated with an increased cancer risk. As the determination of optimal surveillance strategies is hampered by wide ranges in cancer risk estimates and lack of data on cancer-related mortality, we assessed cancer risks and mortality in a large cohort of patients with PJS.
Methods Dutch PJS patients were included in this cohort study. Patients were followed prospectively between January 1995 and July 2009, and clinical data from the period before 1995 were collected retrospectively. Data were obtained by interview and chart review. Cumulative cancer risks were calculated by Kaplan–Meier analysis and relative cancer and mortality risks by Poisson regression analysis.
Results We included 133 PJS patients (48% males) from 54 families, contributing 5004 person-years of follow-up. 49 cancers were diagnosed in 42 patients (32%), including 25 gastrointestinal (GI) cancers. The median age at first cancer diagnosis was 45 years. The cumulative cancer risk was 20% at age 40 (GI cancer 12%), increasing to 76% at age 70 (GI cancer 51%). Cumulative cancer risks were higher for females than for males (p=0.005). The relative cancer risk was higher in PJS patients than in the general population (HR 8.96; 95% CI 6.46 to 12.42), and higher among female (HR 20.40; 95% CI 13.43 to 30.99) than among male patients (HR 4.76; 95% CI 2.82 to 8.04). 42 patients had died at a median age of 45 years, including 28 cancer-related deaths (67%). Mortality was increased in our cohort compared to the general population (HR 3.50; 95% CI 2.57 to 4.75).
Conclusions PJS patients carry high cancer risks, leading to increased mortality. The malignancies occur particularly in the GI tract and develop at young age. These results justify surveillance in order to detect malignancies in an early phase to improve outcome.
- Peutz–Jeghers syndrome
- cancer
- mortality
- risk
- cancer epidemiology
- cancer susceptibility
- cancer syndromes
Statistics from Altmetric.com
Significance of this study
What is already known about this subject?
Peutz–Jeghers syndrome (PJS) is characterised by an elevated cancer risk.
The malignancies associated with PJS include gastrointestinal cancers, breast cancer and gynaecological tumours.
What are the new findings?
The cumulative cancer risk is 76% at age 70, which is lower than reported in a previous meta-analysis.
The cancer risk is predominantly elevated in female PJS patients.
PJS is associated with a significant increased mortality compared to the general population (HR 3.5), predominantly due to the elevated cancer risk.
How might it impact on clinical practice in the foreseeable future?
These results can help in clinical management and counselling of PJS patients.
These results justify surveillance of PJS patients in order to improve outcome.
We have formulated a surveillance recommendation to detect gastrointestinal, breast and gynaecological cancers in an early phase.
Introduction
Peutz–Jeghers syndrome (PJS) is a rare autosomal dominant inherited disorder characterised by gastrointestinal hamartomatous polyposis and mucocutaneous pigmentations.1 Germline mutations in the STK11 gene (serine threonine kinase 11, also known as LKB1 gene) cause PJS and genetic testing is now widely available.2 3 With the currently available techniques a pathogenic STK11 germline mutation can be detected in 80–94% of families with the PJS phenotype.4 5
It has been recognised that PJS is associated with an increased risk for the development of gastrointestinal (GI) and extra-GI malignancies such as breast cancer and gynaecological carcinomas. The malignancies associated with PJS have been reported to occur at a young age,6 7 which is consistent with the identification of the STK11 gene as a tumour suppressor gene.8 Besides, the spectrum of PJS includes the frequent occurrence of tumours that are considered to be benign or with low malignant potential such as Sertoli tumours of the testis and certain sex cord tumours of the ovarium.9 10
Since PJS is a rare disorder, it is difficult to assess cancer risks properly. So far, only large collaborative studies evaluating heterogeneous groups of patients11–13 or small cohort studies14–21 have been conducted, leading to a wide range in reported cancer risk estimates.7 The wide range in risk estimates hampers the development of optimal surveillance strategies and adequate counselling of PJS patients. In addition, data on the mortality in PJS patients are lacking. Therefore the aim of this study was to evaluate a large homogenous cohort of PJS patients in order to determine (1) cumulative cancer risks, (2) the relative cancer risk compared to the general population, and (3) the mortality rate in PJS patients compared to the general population.
Methods
All PJS patients from two Dutch academic hospitals were included in this cohort study between 1995 and July 2009. After informed consent, we included patients without selection for medical history. All patients had a definite diagnosis of PJS, defined by diagnostic criteria recommended by the WHO (box 1),22 a proven STK11-mutation, or both. The study was approved by the Institutional Review Board of both participating hospitals.
Box 1 Diagnostic criteria for Peutz–Jeghers syndrome (PJS) recommended by the WHO22
Positive family history of PJS, and
1. Any number of histologically confirmed PJS polyps*, or
2. Characteristic, prominent, mucocutaneous pigmentation.
Negative family history of PJS, and
1. Three or more histologically confirmed PJS polyps, or
2. Any number of histologically confirmed PJS polyps and characteristic, prominent, mucocutaneous pigmentation.
↵* Histology PJS polyps: A central core of smooth muscle that shows tree-like branching, covered by the mucosa native to the region which is heaped into folds producing a villous pattern.
Patient information was obtained by interview and chart review. The following data were collected per patient: date of birth, establishment of PJS diagnosis, STK11 mutation status, family history of PJS, diagnosis and characteristics of cancer(s) and date and cause of death. Cancer characteristics that were recorded included date of diagnosis, tumour type and origin, tumour invasion (carcinoma in situ (CIS), or invasive carcinoma), as well as data on confirmation (anamnestic, medical record or histology), and presentation (surveillance or complaints/other). Cause of death was classified as cancer-related, intussusception-related or other. Patients were followed prospectively between January 1995 and July 2009, and clinical data from the period before 1995 were collected retrospectively. In addition, family pedigrees were traced backwards and laterally as far as possible, and data of deceased family members fulfilling the diagnostic criteria for PJS were also collected.
Statistical analysis
Data were analysed using the SPSS 15.0 statistical software and the R-2.6.0 statistical package for Windows. Cumulative age-specific cancer risks were calculated with the Kaplan–Meier method and the Cox proportional hazards model. Cumulative cancer risks were determined for any cancer (overall cancer risk) and GI cancer. The following cancers were classified as GI cancer; colorectal, small intestinal (including ampullary cancers), stomach, esophageal, pancreatic and biliary cancers, as well as adenocarcinomas from the digestive tract (not otherwise specified). In the case of multiple primary tumours only the first malignancy contributed to the calculation of the overall cumulative cancer risk. Second primary malignancies were included in the analysis of the cumulative GI cancer risk, yet in patients with two GI cancers only the first event contributed.
The relative cancer risk, that is the ratio of cancer risk in the Dutch PJS cohort compared to the cancer risk in the Dutch general population, was calculated for the period between 1960 and 2009 by Poisson regression analysis. Although PJS patients can be considered at risk for cancer (and death) from birth, person-years at risk were calculated from the age of 5 years until the date of cancer diagnosis, date of death, date of last contact or the closing date of the study. We calculated person-years at risk from the age of 5, since establishment of the PJS diagnosis before that age can be difficult. As a result, recording of events (death or cancer incidence) in the first years of life is incomplete, as reflected by the fact that there was no perinatal mortality in our cohort. To adjust for possible differences in associated cancer risk between males and females, different age groups, and different time periods, the number of person-years were calculated according to sex and subdivided into 5-year age groups and 5-year calendar periods. Gender, age, calendar period and tumour-specific cancer incidence rates of the general Dutch population were derived from the Comprehensive Cancer Centre South (1960–2007). These data are representative for The Netherlands.23 Incidence rates for 2007 were assumed to be representative for 2008 and 2009. For the calculation of the relative cancer risk, only the first event contributed in case of multiple primary tumors. Since our cohort was too small to calculate relative cancer risks according to tumour origin (ie, GI cancer), we determined the distribution of tumour origin according to age in the PJS cohort and compared this to the distribution in the general population by the Fisher's exact test (not adjusted for calendar period). For this comparison, age was categorised according to 25th and 75th percentile of age at first cancer in the PJS cohort: 5–34 years, 35–54 and 55 or older. Cancers diagnosed in the PJS cohort before 1960 were not included in these analyses as data on cancer incidence in the general population were not available for the period before 1960. Gynaecological cancer was defined as cancer originating in the cervix, uterus and adnexa/ovaries.
Mortality in the PJS cohort was compared to mortality in the general population for the period between 1880 and 2009 by Poisson regression analysis in a similar manner as described above. Data on mortality in the general Dutch population between 1880 and 2006 were derived from the Human Mortality Database (University of California, Berkeley, USA, and Max Planck Institute for Demographic Research, Germany).24 The number of deaths in the general population in 2006 was assumed to be representative for 2007, 2008 and 2009. We calculated overall relative cancer and mortality risks (HRs), and also compared relative cancer and mortality risks between males and females, between the period before and after 1970, and between individuals younger and older than 45 years by adding interaction terms.
Results
Study population
A total of 133 PJS patients were included, contributing to a total of 5004 person-years of follow-up (including 1400 person-years of prospective follow-up). These included 64 males (2687 person-years) and 69 females (2317 person-years). At the closing date of the study, two patients (1%) had been lost to follow-up, 42 patients (32%) had died at a median age of 45 (range 3–76 years), and the median age of the 89 patients (67%) still alive was 34 years. Baseline characteristics of the 133 included patients are shown in table 1. They came from 54 different families, including the original Peutz kindred.1 25 Probands generally came under medical attention because of the combination of pigmentations and complications of polyposis, predominantly abdominal pain. None of the probands presented with cancer at the first consultation. One hundred patients in the cohort had a family history of PJS, and the number of patients affected with PJS per family ranged from 2 to 24 (median 3, including proband). STK11 mutation analysis was performed in 80 patients (60%), and a pathogenic germline mutation was detected in 77 patients (96% of patients tested). A total of 81 patients underwent some kind of surveillance.
Cancer spectrum
Forty-nine cancers were diagnosed in 42 of the 133 included patients (32%), including seven patients diagnosed with two primary carcinomas. The characteristics of the 49 malignancies, including 25 GI cancers, six gynaecological cancers and six breast cancers, are shown in table 2. The GI cancers included seven colorectal, six small intestinal, four gastric, three pancreatic and two biliary cancers as well as three adenocarcinomas from the digestive tract not further specified (table 3). The median age at first cancer diagnosis was 45 years (range 15–76 years), and the median age at first GI cancer diagnosis was 42 years (range 15–76 years). There was no difference in cancer incidence between index cases (probands) and their relatives (p=0.46), or between patients from the two participating hospitals (p=0.52). Nineteen of the 42 patients diagnosed with cancer participated in a surveillance programme; in six of the 42 patients (14%) the malignancy was discovered during surveillance, including three carcinomas in situ (one colorectal and two gastric lesions). Twelve of the 19 patients under surveillance developed cancer in an organ not being under surveillance (ie, lung cancer), and one male patient undergoing surveillance of the GI tract was diagnosed with liver metastasis of an adenocarcinoma of unknown primary origin. Another two cancers, both located outside the GI tract and diagnosed in 1960 and 1964 respectively, were solely based on anamnestic data and could not be confirmed histologically or by a medical report.
Cumulative cancer risk
The Kaplan–Meier estimate for the cumulative cancer risk was 20%±5% at age 40 (GI cancer 12%±4%), 36%±6% at age 50 (GI cancer 21%±5%), 58%±7% at age 60 (GI cancer 35%±8%), and 76%±8% at age 70 (GI cancer 51%±10%) (figure 1). The cumulative cancer risk was higher for females than males with PJS (HR 5.14; 95% CI 1.63 to 16.2, p=0.005), but univariate analysis (log-rank) showed no difference in cumulative GI cancer risk between males and females (p=0.845). There was no significant difference in cumulative cancer risk between sporadic PJS cases and patients with a family history of PJS (HR 0.56; 95% CI 0.17 to 1.85, p=0.34), nor did mutation status affect cancer risk (HR 1.01; 95% CI 0.11 to 9.03, p=0.99).
Relative cancer risk
Poisson regression analysis showed that the overall relative cancer risk was significantly higher in PJS patients than in the general population (HR 8.96; 95% CI 6.46 to 12.42, p<0.001), with a higher risk (p<0.001) in females (HR 20.40; 95% CI 13.43 to 30.99) than in males with PJS (HR 4.76; 95% CI 2.82 to 8.04) as compared to the female and male general population, respectively. There was no relative cancer risk difference (p=0.28) between the period before (HR 16.50; 95% CI 5.32 to 51.14) and after 1970 (HR 8.60; 95% CI 6.11 to 12.10), but there was a trend (p=0.08) for a higher relative cancer risk in PJS patients <45 years (HR 12.88; 95% CI 7.89 to 21.02) than in PJS patients ≥45 years (HR 7.20; 95% CI 4.65 to 11.16). Table 4 demonstrates that GI cancers account for a significantly larger proportion of all cancers in our PJS cohort than in the general population, especially among males and younger patients.
Mortality
The cause of death of the 42 deceased patients is depicted in table 5. Most patients died as a result of cancer (67%) or bowel intussusception (19%). All intussusception-related deaths occurred before 1970. The median age at death was 45 years (range 3–76 years). The youngest seven deceased patients died as a result of an acute intussusception at ages between 3 and 20 years, and the youngest age at cancer-related death was 30 years. There was a clear excess mortality risk of 250% in the PJS cohort compared to the general population (HR 3.50; 95% CI 2.57 to 4.75, p<0.001). Table 5 shows that the mortality excess tended to be higher (borderline significance, p=0.076) among female than male PJS patients. There was no relative mortality risk difference between the period before and after 1970 or between PJS patients <45 years and ≥45 years, as compared to the general population.
Discussion
This prospective cohort study demonstrates that PJS patients carry a markedly elevated cancer risk at young age, concerning primarily cancers in the GI tract. The life-time cumulative cancer risk is more than 76% in PJS patients, and higher in females than in males, but independent of family history and STK11 mutation status. The relative cancer risk adjusted for age, sex and calendar period is nearly 10 times higher than in the general population, and the relative cancer risk was most pronounced in females suffering from PJS. The fact that cancer risks are higher in female PJS patients than in male patients can largely be explained by the additional risk of breast cancer and gynaecological cancers, as there was no difference in cumulative GI cancer risk between males and females. In addition to the elevated cancer risk we also demonstrated an increased mortality in PJS patients compared to the general population, with a trend for a larger mortality excess among female PJS patients than in males with PJS. The increased mortality can in part be explained by the elevated cancer risk. Apart from cancer, acute bowel intussusception was, at least before 1970, another important cause of death. Our results suggest that surveillance of PJS patients may prolong life expectancy and improve outcome of these patients, by early detection of carcinomas and timely removal of hamartomas in order to prevent intussusception.
The presented elevated cancer risk is in line with previous reports on the increased cancer risk associated with PJS.6 11–21 27–33 The cumulative cancer risks in relation to age were similar in our study as in a previous study,12 that is 20 versus 17% at age 40, 36 versus 31% at age 50, 58 versus 60% at age 60 and 76 versus 85% at the age of 70 years. However, in general, the relative and cumulative cancer risks in our study are lower than the previously reported relative cancer risk up to 18 and life-time cumulative cancer risk up to 93%.6 12 13 15 19 29 This might be explained by differences in period of follow-up or the use of other inclusion criteria. In the present study patients were included on the basis of the WHO criteria, whereas most other studies used clinical criteria as described by Giardiello and colleagues.15
Our study has some important advantages over the previously performed studies. We describe a unique large pedigree-based cohort of Dutch patients with thorough case ascertainment and a substantial prospective period of follow-up. In contrast, previous large studies concern a meta-analysis (with a similar number of person-years of follow-up as in our study)6 and multinational collaborations, retrospectively evaluating heterogeneous PJS cohorts.12 13 29 Collaborative studies may introduce bias in calculating relative cancer risks since cancer incidences vary between countries, whereas we compared Dutch PJS patients exposed to similar environmental factors, to the Dutch general population. Another advantage of our study is that the relative cancer risks calculated have not only been adjusted for age and sex like in the previous studies, but also for calendar period. Moreover, this is the first study to specifically report on mortality and cause of death among PJS patients.
However, this study also has some limitations. First, selection bias may have resulted in overestimated cancer risks, as particularly patients with most severe phenotypes may have been included. Nevertheless, patients were included systematically regardless of their medical history to minimise this form of bias, and there was no difference in cancer incidence between probands and relatives making ascertainment bias less likely. In addition, referral bias might have led to overestimation of cancer risks, if only patients with a severe course of the disease (including intussusception and cancer) have been referred to our specialised centres. Furthermore, detailed data on surveillance in our cohort are missing since there has not been a nationwide implemented surveillance strategy for PJS patients in The Netherlands until recently. One might postulate that surveillance could have increased the cancer incidence in our population, as six (12%) of the 49 cancers in our cohort, including three intramucosal neoplasias, were diagnosed during surveillance in asymptomatic patients. Yet, the development of a symptomatic invasive carcinoma would have been a matter of time in these young patients (26, 30, 35, 37, 49 and 61 years old, respectively), and therefore these cases were included in our risk analyses.
Another pitfall in the calculation of the cancer risk in PJS is a phenomenon called pseudo-invasion, which is epithelial displacement through the muscularis mucosae, mimicking an invading carcinoma.34–36 This phenomenon has been observed in approximately 10% of small-bowel polyps in one study,37 and can be distinguished from invasive carcinoma by the lack of cytological atypia. On histological revision, we found one case of pseudo-invasion of the small bowel in our cohort, which has been excluded from further cancer-risk analyses.34 However, no tissue was available from two small intestinal cancers (table 2), thus pseudo-invasion could not be ruled out in these two cases.
The exact mechanism of carcinogenesis in PJS remains to be established. One unresolved question, important in light of surveillance, is whether the malignancies in the stomach, small intestine and colorectum originate from the hamartomas or from coexisting adenomas or otherwise normal appearing mucosa.38 39 Although several studies have reported a hamartoma–adenoma–carcinoma sequence,17 40 41 other facts contradict this theory. For example, the number of polyps decreases with advancing age, whereas the cancer incidence increases with advancing age.6 33 42 Furthermore, the location of GI cancers in PJS patients does not always correlate with the location of the hamartomas,39 as we also demonstrated in this series; more colorectal cancers (n=7) were diagnosed than small intestinal cancers (n=6), whereas the small intestine is the preferential localisation for the hamartomas.
To answer the question whether or not hamartomas are pre-malignant and to gain more insight into PJS-related carcinogenesis, further basal research is required. Although it has been suggested that a second gene locus responsible for PJS might exist,43 44 we believe that future research should focus on the STK11 gene function since STK11 mutations can already be detected in more than 90% of patients (96% in these series) with new available techniques.4 5 The development of hamartomas and malignancies might be independent stromal and epithelial processes,8 which complicates the elucidation of the molecular mechanisms underlying STK11-associated carcinogenesis. The exact role of STK11 in the carcinogenic pathway is still unclear, but upregulation of mTOR signalling seems to be an important step as mTOR inhibitors have been shown to reduce tumour burden in mouse models.45 46 Elucidating the molecular background of cancer susceptibility in PJS patients might reveal therapeutic options.
Nevertheless, the cancer risks observed in our study are very high and come close to other high-risk conditions for which surveillance has been recommended.47 48 Although the wide tumour spectrum in PJS makes screening for cancer a difficult task in the individual patient, the high cancer risks justify surveillance of PJS patients. Several surveillance recommendations for PJS have previously been published.49–59 On the basis of the results presented here, our previous review on the elevated cancer risk in PJS,7 and the elevated intussusception risk in PJS patients (submitted), we have also formulated a surveillance recommendation in collaboration with a national working group (table 6).7 The main differences between this surveillance recommendation and previously published guidelines have been discussed earlier.7 Our surveillance recommendation is solely based on expert opinion, since no controlled trials have been published on the effectiveness of surveillance in PJS. The optimal surveillance strategy remains to be established in prospective trials.
In conclusion, PJS patients carry a markedly elevated cancer risk with cancers in particular affecting the GI tract. Female patients are furthermore at high risk for the development of gynaecological tumours and breast cancer. The increased cancer risk, combined with the elevated intestinal intussusception risk, leads to an increased mortality. Although the benefits of surveillance remain to be established, surveillance seems justified. The effect of surveillance on the cancer and intussusception incidence, outcome and survival, as well as cost-effectiveness, will have to be established in prospective trials.
Acknowledgments
The authors would like to thank Dr F Santegoets, Erasmus MC University Medical Center, for his help in the database development.
References
Footnotes
↵* Histology PJS polyps: A central core of smooth muscle that shows tree-like branching, covered by the mucosa native to the region which is heaped into folds producing a villous pattern.
Competing interests None.
Ethics approval This study was conducted with the approval of the The institutional Review Board of both participating hospitals.
Provenance and peer review Not commissioned; externally peer reviewed.