To enhance the precision of the risk estimate for breast cancer in hyperprolactinemia patients by collecting more data and pooling our results with available data from former studies in a meta-analysis.
Population-based cohort study and meta-analysis of the literature.
Using nationwide registries, we identified all patients with a first-time diagnosis of hyperprolactinemia during 1994–2012 including those with a new breast cancer diagnoses after the start of follow-up. We calculated standardised incidence ratios (SIRs) as a measure of relative risk (RR) using national cancer incidence rates. We performed a meta-analysis, combining data from our study with data in the existing literature.
We identified 2457 patients with hyperprolactinemia and 20 breast cancer cases during 19 411 person-years of follow-up, yielding a SIR of 0.99 (95% CI 0.60–1.52). Data from two additional cohort studies were retrieved and analyzed. When the three risk estimates were pooled, the combined RR was 1.04 (95% CI 0.75–1.43).
We found no increased risk of breast cancer among patients with hyperprolactinemia.
Human prolactin (PRL), secreted from the pituitary gland, is central in mammary gland development and lactogenesis, and the PRL receptor is expressed on the mammary gland (1, 2). PRL contributes to initiation, progression, and invasion of breast cancer in both rodent models and humans (2). Whether these effects are mediated by endocrine or autocrine/paracrine mechanisms is uncertain, but epidemiological data suggest that high circulating PRL levels among healthy women are associated with increased breast cancer risk (3). This association has been found to be strongest for postmenopausal women with estrogen-receptor-positive breast cancer (3).
Hyperprolactinemia is usually caused by a PRL-producing adenoma. It is most frequently diagnosed in premenopausal women with menstrual disturbances, infertility, and galactorrhea. The first-line treatment (dopamine agonists) normalizes PRL levels and reduces tumor size in most patients (4). However, a delay in diagnosis or treatment failure may cause prolonged exposure to elevated PRL levels. It is also not unusual for hyperprolactinemia with no or mild symptoms to be left untreated (4). It is thus relevant to assess the risk of breast cancer in women with hyperprolactinemia. Two earlier studies, comprising 1169 and 1342 patients, did not report an increased breast cancer risk in patients with this condition (5, 6). However, the low incidence of breast cancer in mainly young premenopausal women indicates the imprecision of the studies' risk estimates and the need for further investigation.
We therefore conducted a population-based cohort study to examine the risk of breast cancer among Danish patients diagnosed with hyperprolactinemia during 1994–2010. As the number of young women with breast cancer is generally low, we enhanced the precision of the risk estimate by pooling our results with available data from the literature in a meta-analysis.
Subjects and Methods
Population-based cohort study in Denmark
Denmark has tax-funded universal health care, with equal access to hospitals and primary medical care for all residents. The health care system functions in a setting with a tradition of registering most major life events and contacts with the health system, including births, deaths, hospital visits, and dispensed prescriptions (7). Individual records from all registries are linkable using a unique identifier, which is assigned at birth or immigration and encodes sex and age. The Danish Civil Registration System tracks vital status and place of residence (8).
In 1994, the specific code for hyperprolactinemia in the International Classification of Diseases, 10th Revision (ICD-10) was introduced in Denmark. We therefore used the Danish National Patient Registry (DNPR) (9) to identify all patients with a first-time diagnosis of hyperprolactinemia from 1994 to 2012. The DNPR also provided information on comorbidities included in the Charlson comorbidity index (10), which has high validity in the DNPR (11).
For hyperprolactinemia patients, we obtained data on cancer via linkage to the Danish Cancer Registry (12). Patients with a previous breast cancer diagnosis before the start of follow-up (date of hyperprolactinemia diagnosis) were excluded. Each patient was followed from date of hyperprolactinemia diagnosis until date of breast cancer diagnosis, date of death, emigration, or 31st December 2010, whichever came first.
ICD-10 coding for hyperprolactinemia in the DNPR was validated in a subsample of all patients diagnosed at Aarhus University Hospital. Each patient record was retrieved and reviewed with a special emphasis on etiology, i.e. macroprolactinoma or microprolactinoma, drug-associated hyperprolactinemia, or idiopathic hyperprolactinemia. In addition, information about treatment strategies (dopamine-agonist treatment, pituitary surgery) was scrutinized. Hyperprolactinemia was defined as an elevated PRL level according to laboratory reference values.
We used standardised incidence ratios (SIRs) as a measure of relative risk (RR), calculated as the ratio of observed-to-expected cases of breast cancer in the hyperprolactinemia cohort. The number of expected cases was estimated for the entire Danish population based on national cancer incidence rates by age in 1-year intervals, sex, and calendar year in 1-year intervals. We used Byar's approximation to calculate associated 95% CIs, assuming a Poisson distribution of the observed number of cases. When the observed number was less than ten, we instead used exact 95% CIs. We stratified the SIRs by sex, age at hyperprolactinemia diagnosis (<30, 30–49, 50–69, and >70 years), calendar period of hyperprolactinemia diagnosis (1994–1998, 1999–2003, 2004–2008, and 2009–2012), and length of follow-up (0–6, 7–12, or >12 months).
Systematic review and meta-analysis
To identify published studies on the risk of breast cancer in patients diagnosed with prolactinoma/hyperprolactinemia, we conducted a search for all publications in English, Danish, Dutch, Norwegian, and Swedish on the topic (all languages spoken by the authors). The PubMed, Scopus, and Embase databases were searched from 1986 through May 2014. In collaboration with a trained librarian, we constructed a search string focusing on prolactinoma/hyperprolactinemia AND breast neoplasms.
Original studies of two types were eligible for inclusion: cohort studies estimating the RR for breast cancer in patients with hyperprolactinemia/prolactinoma compared with the general population and case–control studies in breast cancer patients estimating the exposure odds ratio for hyperprolactinemia/prolactinoma. Given the low incidence of breast cancer in a hyperprolactinemia/prolactinoma-only cohort, only studies with more than 100 patients were considered. As hyperprolactinemia/prolactinoma is not prevalent in breast cancer, only case–control studies with more than 100 cases were considered.
Evidence review and analysis
Initial selection of studies by title and abstract was performed by one reviewer (M Bengtsen). Selected studies were retrieved for closer scrutiny by two reviewers (M Bengtsen and J O L Jørgensen) and disagreements were resolved by consensus. Retrieved articles were screened using a gauge for judging whether they met the inclusion and exclusion criteria. RR estimates were pooled in a random effects model by default. A fixed effect analysis had to be performed since less than five studies were available, as between-study variability cannot be estimated reliably in that instance.
The study was approved by the Danish Data Protection Board (record no 1-16-02-1-08).
Population-based cohort study in Denmark
We identified 2457 patients with hyperprolactinemia, 2130 (78%) women and 327 (13%) men. Table 1 shows the age distribution at diagnosis, calendar period of diagnosis, and Charlson comorbidity index scores.
Distribution of sex, age, and calendar period of diagnosis and comorbidity among patients with a diagnosis of hyperprolactinemia, Denmark, 1994–2010.
|Age at hyperprolactinemia diagnosis (years)|
|Calendar period of hyperprolactinemia diagnosis|
|Charlson comorbidity index score at hyperprolactinemia diagnosis|
Twenty breast cancer cases were observed during 19 411 person-years of follow-up, compared with 20.3 cases expected in the general population, yielding a SIR of 0.99 (95% CI 0.60–1.52). Table 2 shows SIRs stratified by sex, age at hyperprolactinemia diagnosis, and calendar year of hyperprolactinemia diagnosis. There were no cases of breast cancer among men, and 16 of the 20 breast cancer cases among women occurred in females diagnosed with hyperprolactinemia at ages 30–49. Two breast cancer cases were diagnosed in the first year after the cohort-defining diagnosis of hyperprolactinemia; 18 cases of cancer were diagnosed more than 12 months after this diagnosis (range 16 months–16 years). The absolute 5-year breast cancer risk in the cohort was low: 0.34 (95% CI 0.68–1.15).
Observed and expected cases of breast cancer and standardised incidence ratios (SIRs) for breast cancer with 95% CIs among patients with a diagnosis of hyperprolactinemia overall and by sex, age, calendar time, and comorbidity, Denmark, 1994–2010.
|Characteristic||Observed cases||Person-years||Expected cases||SIR (observed/expected)||95% CI|
|Age at hyperprolactinemia diagnosis (years)|
|Calendar period of hyperprolactinemia diagnosis|
|Length of follow-up (months)|
Hospital records for patients diagnosed with hyperprolactinemia: clinical characteristics and validity of the ICD-10 code
We attempted to retrieve the medical records of all 272 patients from Aarhus University Hospital with a first-time diagnosis of hyperprolactinemia (ICD-10 code E22.1) between 1994 and 2010, including both inpatients and outpatients. Records were available for 264 patients, of whom 222 (84%) were women. Among the 264 patients, three men and 15 women (7%) were misclassified in the DNPR because normal PRL levels were recorded in their medical record at time of hyperprolactinemia/prolactinoma diagnosis; hyperprolactinemia was biochemically confirmed in 229 (92%) of the remaining 246 patients. Among women with a validated diagnosis (n=207), an adenoma was detected on imaging in 70% and hyperprolactinemia was idiopathic or drug-associated in 30%. The most common cause of hyperprolactinemia in males was a macroadenoma (62%). The positive predictive value for the ICD-10 code for hyperprolactinemia in the DNPR, based on patients with retrievable records, was 0.93 (95% CI 0.89–0.96). A more conservative estimate (persons with non-retrievable records counted as unconfirmed) yielded a positive predictive value of 0.87 (95% CI 0.82–0.91).
Meta-analysis and literature review
Our initial search yielded 272 publications, of which 48 were excluded since they were not written in either English, Dutch, or Scandinavian languages; none of the 48 publications, however, were considered to contain data from cohort studies (see Fig. 1 for flow-chart). Two hundred and twenty-two publications were excluded based on title and abstract (Fig. 1). The two remaining studies were eligible for inclusion, one from The Netherlands and the other from Sweden (5, 6). Both studies were based on national databases with linkage to cancer registries. The Dutch study included 1342 females with dopamine-agonist-treated hyperprolactinemia. The Swedish study included 969 hyperprolactinemia patients (668 females). Both studies reported a RR for breast cancer close to 1 (1.07 (95% CI 0.5–2.03) and 1.09 (95% CI 0.6–1.99) respectively). When we pooled these two estimates together with our study's estimate in a meta-analysis (Fig. 2), the combined RR was 1.04 (95% CI 0.75–1.43). No clear heterogeneity was found (I2 0%).
Our population-based cohort study showed no increased risk of breast cancer in patients with hyperprolactinemia. This finding is in line with two previous studies also based on registry data (5, 6). Pooling these three studies in a formal meta-analysis also yielded no evidence of increased breast cancer risk.
The present study was based on linkage of clinical databases and focused on patients with an ICD-10 code for hyperprolactinemia in the DNPR. To assess the validity of the diagnostic code for hyperprolactinemia, we scrutinized medical records for a subsample of patients. The diagnosis turned out to be correct in 93%. A limitation of our study is the lack of laboratory data. We were therefore unable to relate the PRL level to breast cancer risk. Although stratification by PRL levels would be illuminating, two considerations are important. First, in men, who generally present with much higher PRL levels than women, no increased risk was shown. Second, our overall RR was almost exactly 1.0, which makes effect modification unlikely.
There is substantial evidence from both in vitro and animal studies that PRL plays a role in breast cancer development (13). PRL has mitogenic actions in breast cells, inhibits apoptosis, and promotes angiogenesis in breast cancer cell lines, in addition to stimulating the development of metastases in animal models (13). Moreover, data from a large prospective nested case–control study within the Nurses' Health Study (NHS) revealed a positive association between circulating PRL levels and breast cancer risk in both premenopausal (14) and postmenopausal women (15), with RRs of 1.5 (95% CI 1.0–2.5) and 1.3 (95% CI 1.0–1.8) respectively.
The explanation for the discrepancy between the results from the NHS, which included women without evidence of hyperprolactinemia, and those obtained for patients with diagnosed hyperprolactinemia is uncertain. One possible explanation could be that prolonged hyperprolactinemia induces hypogonadism including suppressed estradiol (and androgen) levels, which may reduce the risk for breast cancer, counterbalancing a tumorigenic effect of PRL per se. Moreover, it could be speculated that treatment of hyperprolactinemia with ensuing prolonged suppression of PRL levels outweighs the breast cancer risk associated with pre-treatment hyperprolactinemia. At the same time, it is conceivable that women with hyperprolactinemia have fewer pregnancies, delayed parity, and a higher use of oral contraceptives and combined hormone therapy, all of which would increase the risk of breast cancer. Finally, it remains unknown if the prevalence of other well-known risk factors for breast cancer – such as genetic predisposition, predisposing benign breast conditions, postmenopausal obesity, physical inactivity, and alcohol intake – is lower among women with hyperprolactinemia.
In conclusion, our study shows that a diagnosis of hyperprolactinemia is not associated with increased risk of breast cancer. However, this does not constitute evidence against an active role of PRL in the development and clinical course of breast cancer, which merits future research. Our data may be of use in the treatment and care of women with hyperprolactinemia, including counseling regarding risk prediction of breast cancer.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
This work was supported by the Program for Clinical Research Infrastructure established by the Lundbeck and the Novo Nordisk Foundations, the Danish Cancer Society (R73-A4284-13-S17), and the Aarhus University Research Foundation.
ThygesenSChristiansenCChristensenSLashTSorensenH. The predictive value of ICD-10 diagnostic coding used to assess Charlson comorbidity index conditions in the population-based Danish National Registry of Patients. BMC Medical Research Methodology20111183. (doi:10.1186/1471-2288-11-83).