Secondary Logo

Journal Logo

Breast Cancer: Original Article

Mastitis and the Risk of Breast Cancer

Lambe, Matsa,b; Johansson, Anna L. V.a; Altman, Daniela; Eloranta, Sandraa

Author Information
doi: 10.1097/EDE.0b013e3181adbb1e
  • Free


Inflammatory processes may promote the development of cancer by affecting cell survival, implantation, and growth.1 Inflammation, induced by infections or irritants, has been linked to several forms of cancer,2 contributing to an estimated 1 in 4 of all cancer cases worldwide.3 While mesotheliomas, bladder, cervical, gastric, esophageal, and colorectal cancers have convincingly been shown to arise from sites of inflammation,4 little is known about the role of inflammation in the etiology of breast cancer. Evidence for a possible link comes from epidemiologic observations suggesting that nonsteroidal anti-inflammatory drugs (NSAIDs) protect against the development of breast cancer.5 Also, experimental models have provided some evidence that mammary carcinogenesis is inhibited by NSAIDs.6

Puerperal or lactational mastitis refers to the inflammation of the breast in connection with pregnancy or breast-feeding, and affects as many as 20% of all women within 6 months period after delivery.7 It is most often caused by milk stasis due to inefficient milk removal. The much less common condition of nonpuerperal mastitis refers to inflammatory lesions of the breast unrelated to pregnancy and breast-feeding. While infections play a role in the pathogenesis of both types of mastitis, some cases are completely aseptic.8 Signs and symptoms depend on severity, but often include redness, swelling, tenderness or pain, and systemic symptoms such as fever and abscess formation. The treatment consists of expression of breast milk, skin care, and sometimes antibiotics. Breast abscesses require drainage by aspiration or incision. The risk of mastitis appears to be elevated in women with a previous history of mastitis.9 The aim of this population-based cohort study was to examine possible associations between a history of hospitalization for mastitis and subsequent risk of breast cancer.


Data Sources

The present cohort study was based on data from 6 Swedish population-based registers: the Inpatient Register, the Cancer Register, the Cause of Death Register (held at the National Board of Health and Welfare), the Multi-Generation Register, the Register of the Total Population (held at Statistics Sweden), and the Education Register. Each register includes a national registration number assigned to all Swedish residents that allows record linkage between registers.

The Inpatient Register includes information on hospital admissions and discharges from all public hospitals in Sweden. The database had partial geographic coverage between 1964 and 1986, achieving complete national coverage from 1987. Each Inpatient Discharge record contains the dates of hospital admission and discharge and up to 8 discharge diagnoses, coded according to the International Classification of Diseases (ICD-7 until 1968, ICD-8 from 1968 through 1986, ICD-9 from 1987 through 1996, and ICD-10 thereafter). Women who had been hospitalized with a primary or secondary discharge diagnosis of mastitis (the exposure of interest) were identified in the register using the ICD codes for the respective periods (ICD-7: 689; ICD-8: 675; ICD-9: 678, 689; ICD-10: O910, O911, O9112).

The Swedish National Cancer Register was established in 1958. All newly diagnosed tumors must be reported to the Register by both the clinician and the pathologist or cytologist. Close to 99% of all cases are reported, and in 95% of the cases the Register receives 2 notifications. For all years, the register translates newly reported cancers back to ICD-7. We restricted our study to adenocarcinomas of the breast (for ICD-7, code 170 combined with the pathologic anatomic diagnostic [PAD] code 096 for adenocarcinoma).

The Multi-Generation Register is based on index persons born in 1932 and later and residing in Sweden from 1961 onwards. Index persons are connected with their parents, siblings, and children, allowing retrieval of information on reproductive history, including parity and dates of childbirths.

In addition to providing dates of emigration and immigration, the Register of Total Population and Population Changes includes information on changes of residence within Sweden (domestic migration). The Cause of Death Register, established in 1953, is virtually 100% complete; and it records date and primary and contributing causes of deaths occurring in Sweden. Information in the Population Register and the Cause of Death Register was used to follow individuals over time and assess time at risk. The Swedish Education Register holds information on highest attained educational level, and is continuously updated.

Study Population

The original study population was defined within the Multi-Generation Register as women born between 1944 and 1984; it included 2,635,979 women. Women aged 20 years or more, and residents of Sweden in 1964 or later, were allowed to enter the cohort. As a result, no individual contributed person-time before the start of the Swedish Cancer Register and the Inpatient Register (1958 and 1964, respectively). Due to the incomplete national coverage for the Inpatient Register before 1987, dates of entry were postponed to the date from which each woman had been living continuously in counties with complete coverage, by taking domestic migration into account.

We excluded women from the cohort if (1) they had a previous history of multiple births (n = 43,202), (2) had emigrated before start of follow-up (n = 14,640), (3) had more than one recorded diagnosis of mastitis or an elapsed time between diagnoses of more than 12 months (n = 120), (4) had breast cancer at the beginning of follow-up (n = 193) and (5) had inconsistent ICD classifications (n = 252). In addition, we excluded 5 women where the breast cancer diagnosis preceded the mastitis. Since the first signs of breast cancer in rare cases may mimic those of mastitis, 2 women who had been diagnosed with breast cancer within 6 months of a diagnosis of mastitis were excluded. After the exclusions, the final study population comprised 2,577,565 women.

Data From Medical Records

Information of laterality (left or right) of the breast cancer was available in the records of the Cancer Register. In a separate step, information on laterality of the mastitis was obtained via review of medical records of women identified with both mastitis and breast cancer. We were able to retrieve the medical records for 88 of the 106 women identified in the registers with mastitis and a subsequent diagnosis of breast cancer. For one of these women, information was unavailable on laterality of the breast cancer; this woman was excluded from the analyses of laterality. For 5 cases, bilateral mastitis could not be ruled out; these women were retained in the analysis.

The study was approved by the Research Ethics Committee at Karolinska Institutet, Stockholm.

Statistical Analysis

Standard cohort methods were used to analyze the data. The analyzed event was first primary breast cancer during follow-up. Risk time (in person-years) was accumulated from entry into the cohort (date at age 20, date of immigration, 1 January 1964, or date since residing in regions with complete coverage in the Inpatient Register, whichever came last) until a diagnosis of breast cancer or censoring (the date of first emigration since entry, death, or at end of follow-up on 31 December 2004, whichever came first). Age-standardized incidence rates of breast cancer were calculated as number of events per 100,000 person-years using the Swedish female age distribution as standard population, with 95% confidence intervals (CIs) approximated, assuming a Poisson distribution as described by Breslow and Day.10

To estimate the association between mastitis and breast cancer incidence, we used Poisson regression models, yielding incidence rate ratios (IRRs) with 95% CIs as measures of association. We used age as the underlying timescale in all analyses, splitting the follow-up time of each woman into age intervals (20–29, 30–39, 40–49, and ≥50 years) and including age as a covariate in the models. In addition, one model was estimated adjusting for both age and calendar time as underlying timescales, further splitting the follow-up time of each woman into calendar time intervals (1964–1974, 1975–1984, 1985–1994, 1995–2004). The Poisson model assumes the estimated effects are multiplicative.

Exposure to mastitis was treated as a time-varying exposure and coded as “no” until the time of first mastitis requiring hospitalization, from which time it was coded “yes” throughout follow-up. Highest achieved parity during follow-up was categorized as 0, 1, 2, or ≥3 children. Age at first birth was categorized as <20, 20–24, 25–29, 30–34, and ≥35 years. When adjusting for parity and age at first birth simultaneously, we parameterized the effects so that the reference groups were “1 child” and “age at first birth <20 years,” while still including nulliparous women in the model.11

Likelihood ratio tests were used to assess the effects of exposure and confounder variables on breast cancer rates. The proportional hazard assumption was evaluated by fitting an interaction term between attained age and the mastitis indicator. There was no evidence of nonproportional hazard for the models. Because both the exposure and all the potential confounders were derived from the registers, we assumed that women with no recorded information had no exposure. Thus, there were no women with incomplete data on any of the covariates in the models.

The SAS/STAT software version 9 was used to analyze the data (SAS Institute Inc., Cary NC, USA).


The 2,577,565 women under study contributed 46.3 million person-years to the analyses, with a mean follow-up time of 18 years. At the end of follow-up, 26,519 cases of breast cancer had occurred in the cohort. Of these, two-thirds (65%) were younger than 50 years at diagnosis (mean age of diagnosis: 47 years).

We identified 8411 women with a discharge diagnosis of mastitis, of whom 83% (n = 6990) were primary diagnoses and 17% (n = 1421) were recorded as secondary diagnoses of mastitis. The proportions of primary and secondary diagnoses were identical in women with and without subsequent breast cancer. Among the 8411 women with a mastitis diagnosis, 106 had a subsequent diagnosis of breast cancer.

Of the 8411 women with a history of mastitis, 8184 (97%) had their mastitis diagnosis within 12 months after delivery and 147 (2%) more than 12 months after delivery. The remaining 80 (1%) women were nulliparous at the time of mastitis diagnosis, of which 30 women were pregnant.

The time between mastitis and breast cancer was generally long. Among the 106 women with both conditions, the cancer occurred from 6 months up to 5 years after the mastitis in 8 (7%) women, from 5 up to 10 years in 18 (17%) women, from 10 up to 15 years in 20 (19%), and at least 15 years in 60 (57%) women.

The age-standardized breast cancer incidence rate among women with mastitis was 128 per 100,000 person-years (95% CI = 99–158), and among women without mastitis 103 (101–104) (Table 1).

Incidence Rates of Breast Cancer by Mastitis, Reproductive Characteristics, and Calendar Period

Compared with women with no recorded mastitis, the incidence rate of breast cancer—regardless of laterality—was slightly higher in women with mastitis, yielding an incidence rate ratio of 1.22 (95% CI = 1.01–1.48) after adjustment for age. After further adjustment for parity and age at first birth and calendar period, the association between mastitis and breast cancer remained essentially unchanged (1.23 [1.02–1.49]) (Table 2). In a subanalysis, we excluded 3 women diagnosed with breast cancer in the same breast within 5 years of the mastitis diagnosis. This approach only marginally altered the point estimate (adjusted IRR = 1.20 [0.99–1.46]).

Incidence Rate Ratios of Breast Cancer by Mastitis, Parity, Age at First Birth, and Calendar Period

The risk of breast cancer decreased with increasing parity, with a 24% lower risk among women with 3 or more children as compared with women having one child. The risk of breast cancer increased with increasing age at first birth. Additional adjustment for educational level did not alter these estimates. Including women who experienced multiple events of mastitis did not alter the results, nor did a restriction to parous women (Table 2). In the group of women (n = 87) among whom information on laterality was available for both the mastitis and the malignancy, side of lesions corresponded for 52% (95% CI = 41%–62%).


The possible role of inflammatory response as a precursor to the development of cancer has gained renewed scientific interest in recent years. Converging evidence indicates that inflammatory cells, and the chemokines and cytokines that they produce, act as promoters in the neoplastic process by remodelling the tumor microenvironment. These changes are believed to promote angiogenesis and facilitate the growth, migration, and differentiation, not only of fibroblasts and endothelial cells but also of neoplastic cells.12

Several lines of indirect evidence support the notion that inflammation may influence breast carcinogenesis. First, cytokines (key regulators of inflammation) stimulate the enzymes that convert androsteinedione to estrogens in breast tissue. Exposure to elevated levels of estrogens has been consistently linked to the risk of breast cancer.13 Second, NSAID use has been associated with a reduced risk of breast cancer by 22% overall, and by 28% in frequent users with 10 or more years of exposure.5

In our study, the overall risk of breast cancer was elevated in women who had been hospitalized with a diagnosis of mastitis. However, the absence of a correlation of laterality does not support a causal association. To our knowledge, only one earlier study has attempted to examine the risk of breast cancer in relation to a history of mastitis. In that study, there was an excess risk of dying from breast cancer in a cohort of women with chronic mastitis, although the majority may have suffered from benign fibrocystic disease and not mastitis, according to contemporary definitions. Also, established risk factors for breast cancer were not considered in the analyses.14

The increased overall risk of breast cancer observed in the mastitis group in the present study could be due to residual confounding. Alternatively, one possible biologic explanation is that our findings reflect a shorter cumulative duration of breast-feeding because of nursing problems. Less breast-feeding, in turn, is associated with a higher risk of breast cancer.

Compared with many other countries, the frequency of breast-feeding in Sweden is high. Following a decrease during the 1960s, the proportion of breast-fed children gradually increased. By the mid-1990s, more than 90% of all children were exclusively or partially breast-fed at 2 months.15 There is little documented, systematic knowledge about breast-feeding patterns in women with mastitis. In some women, mastitis-related discomfort disturbs and often precludes prolonged breast-feeding. Results from several studies indicate that one mastitis event predisposes to recurrence.16,17 The experience may also affect breast-feeding patterns in association with subsequent deliveries. In a study that prospectively followed a cohort of breast-feeding mothers, 18% of those who stopped breast-feeding stated mastitis as the primary cause, and mastitis was the third most commonly stated reason for cessation of breast-feeding.18 However, results from an Australian study showed that primiparous women with mastitis were as likely to continue breast-feeding as other women.19 Findings from a New Zealand study indicate that women with mastitis were likely to breast-feed for longer than women without mastitis.20 To our knowledge, no study has specifically examined breast-feeding patterns in hospitalized women, who presumably have more severe mastitis.

Breast-feeding is associated with a reduced risk of breast cancer; in one study combining epidemiologic data on more than 50,000 women, the relative risk of breast cancer was reduced by 4.3% for each year that a woman breastfeeds.21 A protective effect of lactation may be mediated by further terminal differentiation of the breast epithelium, making it more resistant to carcinogenic change. Also, breast-feeding may reduce risk by prolonging the period of anovulation and delaying the reestablishment of the menstrual cycle.

Strengths of the present study included the size of the cohort and the population-based approach. The ascertainment of breast cancer cases in the Swedish Cancer Register can be viewed as essentially complete. The information at hand allowed adjustment for age, calendar time, age at first birth, and parity. Additional adjustment for education level as a proxy for socioeconomic status did not change the risk estimates. No data were available for other established risk factors for breast cancer, such as age at menarche and menopause, breast-feeding, and use of alcohol or exogenous hormones. The main exposure of interest—a history of mastitis—was assessed in the Inpatient Register, restricting the exposure group to hospitalized women. Thus, no information was available on milder cases of mastitis. Assuming that both milder and more severe cases of mastitis may be related to breast cancer risk, this under-ascertainment of women with a past event of mastitis would tend to dilute the observed overall elevated risk estimate of breast cancer toward the null.

In conclusion, while we found evidence of a slight overall increased breast cancer risk in women with a history of mastitis, the combined results of the present study do not support a causal association. Instead, our findings may reflect a shorter cumulative duration of breast-feeding because of pain and discomfort both in association with the pregnancy followed by a mastitis diagnosis, and possibly also in subsequent pregnancies. Other explanations for an indirect association may include yet-to-be identified characteristics of women prone to developing mastitis. Since our analyses were based on a subgroup of patients with mastitis that led to in-hospital care, these findings cannot be immediately extrapolated to women with a history of milder forms of mastitis, a common condition in the postpartum period. Also, the present study was among relatively young women. Future studies in this area would ideally include a wider age-range of women and use a case–control design.


1. Mantovani A, Romero P, Palucka AK, Marincola FM. Tumour immunity: effector response to tumour and role of the microenvironment. Lancet. 2008;371:771–783.
2. Ames BN, Gold LS, Willett WC. The causes and prevention of cancer. Proc Natl Acad Sci U S A. 1995;92:5258–5265.
3. Hussain SP, Harris CC. Inflammation and cancer: an ancient link with novel potentials. Int J Cancer. 2007;121:2373–2380.
4. Balkwill F, Mantovani A. Inflammation and cancer: back to Virchow? Lancet. 2001;357:539–545.
5. Harris RE, Chlebowski RT, Jackson RD, et al. Breast cancer and nonsteroidal anti-inflammatory drugs: prospective results from the Women's Health Initiative. Cancer Res. 2003;63:6096–6101.
6. Steele VE, Moon RC, Lubet RA, et al. Preclinical efficacy evaluation of potential chemopreventive agents in animal carcinogenesis models: methods and results from the NCI Chemoprevention Drug Development Program. J Cell Biochem. 1994;20:32–54.
7. Kinlay JR, O'Connell DL, Kinlay S. Incidence of mastitis in breastfeeding women during the six months after delivery: a prospective cohort study. Med J Aust. 1998;169:310–312.
8. WHO. Mastitis: Cures and Management. Geneva, Switzerland: The World Health Organization; 2000. FCU, CAU/1000.
9. Kinlay JR, O'Connell DL, Kinlay S. Risk factors for mastitis in breastfeeding women: results of a prospective cohort study. Aust N Z J Public Health. 2001;25:115–120.
10. Breslow N, Day N. Statistical methods in cancer research Volume 2. The design and analysis of cohort studies. Lyon: IARC Press; 1987.
11. McKnight B, Cook LS, Weiss NS. Logistic regression analysis for more than one characteristic of exposure. Am J Epidemiol. 1999;149:984–992.
12. Coussens LM, Werb Z. Inflammation and cancer. Nature. 2002;420:860–867.
13. Key TJ, Appleby PN, Reeves GK, et al. Body mass index, serum sex hormones, and breast cancer risk in postmenopausal women. J Natl Cancer Inst. 2003;95:1218–1226.
14. Monson RR, Yen S, MacMahon B. Chronic mastitis and carcinoma of the breast. Lancet. 1976;2:224–226.
15. Swedish National Board of Health and Welfare. Breast Feeding, Children Born 2004. Stockholm, Sweden: Hälsa och sjukdomar; 2006:7.
16. Evans M, Heads J. Mastitis: incidence, prevalence and cost. Breastfeed Rev. 1995;3:65–72.
17. Fetherston C. Risk factors for lactation mastitis. J Hum Lact. 1998;14:101–109.
18. Fetherston C. Characteristics of lactation mastitis in a Western Australian cohort. Breastfeed Rev. 1997;5:5–11.
19. Amir LH, Forster DA, Lumley J, McLachlan H. A descriptive study of mastitis in Australian breastfeeding women: incidence and determinants. BMC Public Health. 2007;7:62.
20. Vogel A, Hutchison BL, Mitchell EA. Mastitis in the first year postpartum. Birth. 1999;26:218–225.
21. Breast cancer and breastfeeding: collaborative reanalysis of individual data from 47 epidemiological studies in 30 countries, including 50,302 women with breast cancer and 96973 women without the disease. Lancet. 2002;360:187–195.
© 2009 Lippincott Williams & Wilkins, Inc.