Statin use and patterns of breast cancer recurrence in the Malmö Diet and Cancer Study.

BACKGROUND: Accumulating evidence suggests that statins have a beneficial effect on breast cancer prognosis. Previous studies have reported a positive association between statin use and breast cancer survival; however, the relationship between statin use and patterns of breast cancer recurrence remains unclear. PATIENTS AND METHODS: We identified all Malmö Diet and Cancer Study (MDCS) participants diagnosed with incident invasive breast cancer between 2005 and 2014. The follow-up period began at breast cancer diagnosis and continued until the first invasive breast cancer recurrence event, death, emigration or the end of the follow-up (June 8, 2020). We estimated incidence rates (IRs) of recurrence and fit Cox regression models to compute crude and adjusted hazard ratios (HRs) with 95% confidence intervals (95% CIs) for disease recurrence to compare post-diagnosis statin users with non-users.
RESULTS: The final study cohort consisted of 360 eligible patients with a median follow-up of 8.6 years. Overall, there were 71 recurrences in 2932 total person-years. According to statin use, there were 14 recurrences in 595 person-years among statin users, and 57 recurrences in 2337 person-years in non-users. Statin use was associated with a reduced risk of breast cancer recurrence (HRadj = 0.88 [95% CI: 0.82-0.96]). Regarding the pattern of recurrence, statin use was associated with a reduced risk of distant recurrence (HRadj = 0.86 [95% CI: 0.80-0.94]) but not loco-regional recurrence (HRadj = 0.97 [95% CI: 0.87-1.08]).
CONCLUSION: In the MDCS, statin use was associated with a reduced risk of distant breast cancer recurrence, whereas no association between statin use and loco-regional breast cancer recurrence was found. This site-based difference in disease recurrence may be explained by statin's inhibition of epithelial-mesenchymal transition.

Introduction

In Sweden, around 8000 women are diagnosed with breast cancer and around 1500 breast cancer-related deaths occur annually, making breast cancer the most prevalent female malignancy and the second leading cause of cancer-related death in the country (https://gco.iarc.fr/). These figures also reflect global breast cancer statistics [1]. Despite screening-based early diagnosis and advanced treatment options, many patients with breast cancer experience a relapse, and 20–30% of breast cancer-related deaths are attributed to disseminated disease rather than the primary tumour [2]. Metastatic dissemination is often preceded by biological reprogramming of the primary cancer cells through an epithelial-mesenchymal transition (EMT) that enables the spread of cancer cells to distant organs [3].

Statins are a group of cholesterol-lowering medications widely prescribed to treat hypercholesterolemia and prevent cardiovascular disease [4]. Statins reduce the levels of low-density lipoprotein cholesterol in circulation by inhibiting de novo cholesterol biosynthesis [4]. Since the early 1990s, mounting scientific evidence has indicated that lipophilic statins have pleiotropic effects that impact the prognosis of many diseases including cancer [[5], [6], [7]]. Preclinical in vitro and in vivo studies indicate that statin treatment potently reduces breast cancer growth and metastasis [[7], [8], [9]]. Three key mechanisms by which statins exert their anti-neoplastic effects are 1) by reducing the synthesis of farnesyl pyrophosphate (FPP) and geranylgeranyl pyrophosphate (GGPP), important by-products of the mevalonate pathway that play significant roles in the prenylation of small guanosine triphosphate-binding signalling proteins [10], 2) by reducing the levels of 27-hydroxycholesterol, a selective estrogen receptor (ER) modulator [11,12], and 3) by suppressing the EMT of primary tumour cells [9,13].

Statin use has consistently been shown to reduce breast cancer-specific mortality and improve overall survival in epidemiological studies [[14], [15], [16]]. However, the prognostic impact of statins in terms of disease recurrence and recurrence patterns is relatively less explored. One population-based observational Danish study reported superior prognostic outcomes for patients with breast cancer who used statins [14]. Later, results from a phase III, endocrine trial including 8010 postmenopausal women with early-stage, hormone receptor-positive, invasive breast cancer reported that concomitant use of cholesterol-lowering medication during endocrine therapy was associated with improved recurrence-free survival and distant recurrence-free interval [17]. Thus, with sizeable preclinical, clinical and epidemiological evidence, we hypothesized that post-diagnosis statin use will have a protective effect in reducing disease recurrence in patients with primary breast cancer. To the best of our knowledge, no study has investigated the impact of statin use on breast cancer recurrence (BCR) in a population-based setting in Sweden. Furthermore, patterns of recurrence according to statin remain poorly understood in general. Thus, this study aimed to assess the association between post-diagnosis statin use and recurrence patterns in a population of patients with breast cancer in Sweden.

Material and methods

The Malmö Diet and Cancer Study

The Malmö Diet and Cancer Study (MDCS) is a population-based prospective cohort study initiated to examine associations between diet and cancer. The study enrolled 17,035 women living in Malmö, Sweden from 1991 to 1996. Prior to their inclusion, the participants answered an extensive questionnaire and research nurses collected anthropometric measures including height, weight, waist and hip circumference. Detailed information regarding MDCS study design, inclusion and exclusion criteria, and data collection have been reported previously [18,19]. Information on incident breast cancer cases and vital status were retrieved annually through record linkage to the Swedish Cancer Registry, the Southern Swedish Regional Tumour Registry and the Swedish Cause of Death Registry [[19], [20], [21]]. Ethical approval was obtained from the Ethical Committee at Lund University (Dnr 427/2007) and all study participants provided their informed consent upon enrolment.

Study population

From the original MDCS cohort, we identified all female participants with incident breast cancer diagnosed between 1991 and 2014 (n = 1240). Patients with pre-1991 breast cancer diagnoses, carcinoma in situ, or bilateral breast cancer and premenopausal patients were excluded from the current study. Furthermore, patients that received neoadjuvant treatment and those with a pre-diagnostic statin prescription were excluded. For the final study population, patients with a breast cancer diagnosis prior to the establishment of the Swedish Prescription Registry in July 2005 [22] were excluded to ensure relevant and complete information on statin use (Fig. 1).

Fig. 1

Flowchart for the study population.

Clinical and pathological information

Information on tumour characteristics, tumour markers, and breast cancer treatments were retrieved from medical records. Additional information on tumour markers was obtained from immunohistochemistry microarray assessments of tumour tissue performed by the Molecular Pathology Laboratory, Malmö University Hospital, Malmö, Sweden. In accordance with the Swedish clinical guidelines, ER status was considered positive if more than 10% of cancer cell nuclei were stained positive.

Endpoints

The primary endpoint of this study was invasive BCR, defined as the time from breast cancer diagnosis to the occurrence of any local, regional, invasive contralateral, second primary breast cancer or distant recurrence. We further analysed three secondary endpoints 1) Loco-regional recurrence defined as the time between breast cancer diagnosis and any local or regional recurrence without metastatic spread, 2) Distant recurrence defined as the time between breast cancer diagnosis and the first evidence of metastatic disease, invasive contralateral or second primary breast cancer and 3) Overall survival (OS) defined as the time from breast cancer diagnosis until death from any cause.

Statin use

Information on statin use was obtained from the Swedish Prescription Registry from July 2005, when the registry was initiated, through 2020. Statin use was handled as a time-varying variable and all patients were assumed non-users of statins at start of follow-up. For a patient to be defined as a statin user in the analyses, the patient had to fill a prescription of statins. One filled statin prescription was assumed to expose the individual from the prescription date through the remaining follow-up time. All patients holding a prescription of statins prior to their breast cancer diagnosis were excluded.

Statistical analyses

Follow-up began at breast cancer diagnosis and continued until any invasive BCR, death, emigration, or the end of available follow-up data on June 8, 2020. Patients with these events were censored at the time of the event. We computed incidence rates for statin exposed/unexposed, and crude/adjusted hazard ratios (HRs) with 95% confidence intervals (95% CIs) of the study endpoints using Cox regression models. Statin use was included in the model as a dichotomous time-varying variable. Two multivariable Cox regression models were fitted. Model 1 was adjusted for age at diagnosis (continuous variable), year of breast cancer diagnosis (continuous variable), body mass index (BMI) (normal weight: BMI < 25 kg/m2, overweight: BMI ≥ 25 and < 30 kg/m2, obese: BMI ≥ 30 kg/m2), ER-status (positive vs negative), node status (N0, N1, N2, N3), Nottingham histological grade (I, II, III) and invasive tumour size (<10 mm, 10 mm–20 mm, > 20 mm). Model 2 included the covariates of model 1 with the addition of surgical procedure (mastectomy or partial mastectomy), intended adjuvant radiotherapy (yes vs no), intended adjuvant chemotherapy (yes vs no) and intended adjuvant endocrine therapy (yes vs no).

To account for the competing risk of death, we used the Fine-Gray competing risk regression model [23] to compute the sub-hazard ratio (SHR) with a 95% CI.

In sensitivity analyses, we tried to limit the misclassification of statin use in the exposed group by requiring a patient to fill five prescriptions of statins to be classified as exposed.

Results

The present study included 360 patients diagnosed with invasive breast cancer in the MDCS during 2005–2014 who were then followed until June 8, 2020. Overall, 71 BCRs occurred over a total of 2932 person-years of follow-up with a median follow-up of 8.6 years (interquartile range (IQR): 5.6–11.1 years). During the follow-up period, 91 patients received at least one statin prescription. The median period of statin use was 5 years (IQR: 2.5–8.6 years) and the median number of statin prescriptions was 30 (IQR: 18–46). Table 1 presents the demographic and tumour characteristics of all patients (n = 360), as well as for statin users (n = 91) and non-users (n = 269). Statin users were older with a median age of 68 years at diagnosis (IQR: 63–73 years) and had a higher waist to hip ratio. Statin use was also associated with lymph node negativity and a higher likelihood of undergoing a partial mastectomy and receiving adjuvant radiotherapy.

Table 1

Demographic and tumour characteristics of patients with breast cancer in the MDCS according to statin use.

	Total	Non-users	Statin users
	n = 360	n = 269	n = 91
Age at diagnosis (years)
50–59	15 (4%)	9 (3%)	6 (7%)
60–69	166 (46%)	125 (47%)	41 (45%)
70 - 79	109 (30%)	78 (29%)	31 (34%)
>79	70 (20%)	57 (21%)	13 (14%)
Nottingham histological grade
I	86 (26%)	62 (25%)	24 (28%)
II	159 (48%)	119 (49%)	40 (46%)
III	86 (26%)	64 (26%)	22 (26%)
Missing	29	24	5
Node status
Negative	239 (71%)	174 (69%)	65 (75%)
Positive	100 (29%)	78 (31%)	22 (25%)
Missing	21	17	4
Tumour size (mm)
<10	57 (17%)	40 (16%)	17 (19%)
10-20	184 (55%)	138 (55%)	46 (53%)
>20	96 (28%)	72 (29%)	24 (28%)
Missing	23	19	4
Estrogen receptor status
Negative	31 (9%)	24 (10%)	7 (8%)
Positive	296 (91%)	218 (90%)	78 (92%)
Missing	33	27	6
Body mass index
<25 kg/m2	195 (55%)	146 (54%)	49 (54%)
25–30 kg/m2	124 (34%)	92 (34%)	32 (35%)
≥30 kg/m2	41 (11%)	31 (12%)	10 (11%)
Waist to hip ratio
≤0.80	213 (59%)	166 (67%)	47 (52%)
0.81–0.85	102 (28%)	72 (21%)	30 (33%)
>0.85	45 (13%)	31 (12%)	14 (15%)
Surgical procedure
Mastectomy	124 (40%)	99 (44%)	25 (30%)
Partial mastectomy	186 (60%)	128 (56%)	58 (70%)
Missing	50	42	8
Endocrine therapy, intended adjuvant
No	103 (30%)	75 (29%)	28 (32%)
Yes	244 (70%)	185 (71%)	59 (68%)
Missing	13	9	4
Chemotherapy, intended adjuvant
No	254 (83%)	195 (83%)	59 (81%)
Yes	53 (17%)	39 (17%)	14 (19%)
Missing	53	35	18
Radiotherapy, intended adjuvant
No	113 (37%)	93 (39%)	20 (27%)
Yes	196 (63%)	143 (61%)	53 (73%)
Missing	51	33	18

Among the 71 BCRs, 14 recurrences occurred during 595 person-years of follow-up among patients exposed to statins and 57 recurrences in 2337 person-years in the unexposed group (IR per 1000 person-years: 23.5 [95% CI 13.9–39.7] and 24.4 [95% CI 18.8–31.6], respectively). As depicted in Table 2, statin users had a significantly lower risk of disease recurrence both in the crude (HRcrude = 0.92; 95% CI 0.87–0.98) and adjusted analyses (HRadj = 0.88; 95% CI 0.82–0.96). Furthermore, Cox regression analyses were performed for loco-regional and distant recurrences separately. The protection seen against any kind of recurrences was particularly driven by reduced distant recurrences (HRadj = 0.86; 95% CI 0.80–0.94) and less protection against loco-regional recurrences was observed (HRadj = 0.97; 95% CI 0.87–1.08).

Table 2

Association between statin use and breast cancer outcome in the MDCS.

Outcome	Exposure	Person-years	Number of events	Incidence rate per 1000 person-years (95% CI)	Crude HR (95% CI)	Model 1a HRadj (95% CI)	Model 2b HRadj (95% CI)
Breast cancer recurrence
	Not exposed to statins	2337	57	24.4 (18.8–31.6)
	Statin exposure	595	14	23.5 (13.9–39.7)	0.92 (0.87–0.98)	0.90 (0.84–0.97)	0.88 (0.82–0.96)
Distant recurrence
	Not exposed to statins	2344	49	20.9 (15.8–27.7)
	Statin exposure	609	11	18.1 (10.0–32.6)	0.93 (0.87–1.01)	0.91 (0.84–0.99)	0.86 (0.80–0.94)
Loco-regional recurrence
	Not exposed to statins	2337	14	5.9 (3.5–9.9)
	Statin exposure	595	6	10.1 (4.5–22.4)	1.01 (0.91–1.09)	0.98 (0.88–1.08)	0.97 (0.87–1.08)
Overall survival
	Not exposed to statins	2494	78	31.3 (25.1–39.0)
	Statin exposure	730	19	18.1 (16.6–40.8)	0.90 (0.83–0.98)	0.93 (0.85–1.01)	0.89 (0.83–0.99)

Model 1: adjusted for age, year of breast cancer diagnosis body mass index, estrogen receptor status, Nottingham histological grade, nodal status and tumour size status.

Model 2: adjusted for Model 1 and surgery, adjuvant endocrine therapy, adjuvant chemotherapy and adjuvant radiotherapy.

In competing risk analyses with death as a competing risk sub-hazard ratios displayed a lower risk of distant recurrences (SHR = 0.84; 95% CI 0.71–0.99) in statin users. However, this association was not observed for loco-regional recurrences (SHR = 0.97; 95% CI 0.82–1.14).

Regarding overall survival, 19 deaths were reported in 730 person-years in the group of patients taking statins (IR per 1000 person-years: 18.1 [95% CI 16.6–40.8]), while 78 deaths were reported in 2494 person-years in the non-exposed group (IR per 1000 person-years: 31.3 [95% CI 25.1–39.0]). As expected, statin users had a reduced risk of death due to any cause in both the crude (HRcrude = 0.90; 95% CI 0.83–0.98) and adjusted (HRadj = 0.89; 95% CI 0.83–0.99) models.

Discussion

Metastatic breast cancer remains a challenge considering the incurable setting, and the prevention of disease recurrence is thus an important area of research and development. Statins are inexpensive, per-oral drugs that have anti-cancer effects with limited side effects. In this prospective, population-based cohort study, we show that post-diagnosis statin use is associated with a significant reduction in disease recurrence, specifically for distant metastases in patients with breast cancer. Thus far, epidemiological studies investigating the role of statins and primary breast cancer risk have yielded mixed results [[24], [25], [26], [27]]. However, studies that assessed the link between statins and breast cancer recurrence have reported consistent protective effects in agreement with our study [14,17,[28], [29], [30]].

The findings from the present study, that statin use significantly reduces distant breast cancer recurrence, support experimental data presenting the antitumour effects of statins [[8], [9], [10]]. By investigating breast cancer recurrence patterns, we aimed to improve the understanding of the molecular mechanisms underlying the beneficial effects of statins. EMT is a cellular differentiation process where epithelial tumour cells are reprogrammed to attain mesenchymal characteristics through molecular and structural alterations. E-cadherin maintains the cell-cell junctions of epithelial cells and the downregulation of this protein along with a concomitant upregulation of N-cadherin expression is associated with metastatic breast cancer outgrowth [31,32]. Experimental studies performed to identify statin-sensitive cancer cell types showed that breast cancer cells with higher expression of mesenchymal cell markers (e.g., vimentin) and/or lower expression of epithelial cell markers (e.g., E-cadherin) are highly sensitive to statin treatment [33,34]. Furthermore, in an elegant study using an ex vivo micro physiological system model of breast cancer metastasis, Beckwitt et al. showed that atorvastatin treatment selectively reduced the outgrowth of breast cancer metastases to the lung and liver [9]. Thus, the findings from these experimental studies may explain the results from this prospective population-based cohort study. Suggesting that exposure to statins could possibly prevent disease recurrence to distant organs.

Our study has some strengths. We used a well-characterized population-based prospective cohort with long follow-up data on site-specific breast cancer recurrence that covers a wide span of early and late events. Second, when studying the association between statin use and breast cancer outcome, disease recurrence is a more reliable outcome than mortality, which tends to have more confounding issues with comorbidities (e.g., cardiovascular disease, diabetes mellitus, obesity) that are strongly related to both the exposure of interest and outcome. Finally, in this study, anthropometric measures are important confounding factors. The availability of reliable anthropometric measures obtained by an experienced research nurse represents an important strength.

This study has some limitations. To identify statin users, we used register-based prescription data, and compliance information was not available. This could induce a misclassification bias, yet in analyses restricted to patients receiving more than five statin prescriptions, the association was not attenuated. The beneficial effect of statins on recurrence observed could be due to a healthy user bias if the prescription of statins were associated with stable health. However, the median time to recurrence in this study among non-users was 8.4 years and 6.5 years among statin users; while the median time to first statin prescription was 5.7 years indicating that a healthy user bias is unlikely to explain the observed results. Another limitation is that the BMI used in the multivariable models was obtained at study entry and not at the time of diagnosis. The optimal approach would have been to use BMI measured at diagnosis or post-diagnostically at the initiation of statin use. Since we observed a statin-specific recurrence pattern (distant vs loco-regional), it would be of great interest to explore if statin use also manifests organ-specific metastasis patterns. However, this was not possible as no such data were available. Several studies have suggested that the protective effect of statins on breast cancer outcome is limited to lipophilic statins [14,35,36]. In this cohort, less than five percent of the patients using statins were prescribed hydrophilic statins and therefore no analyses comparing lipophilic and hydrophilic statins were carried out. Furthermore, our small population limits the interpretation of our results. Yet, it should be noted that a time-varying approach to studying statin exposure within our multivariable models was used to avoid immortal time bias and allow for clear interpretation. Importantly, patients prescribed statins were first exposed to statins when filling a prescription for statins. Thus, we avoided allocating any time prior to treatment initiation as exposure time, which would lead to a false protective effect of statins on breast cancer outcome. In this study, patients using statins before breast cancer diagnosis were excluded to ensure that no modulating effects of pre-diagnostic statin use on tumour characteristics and tumour stage had occurred as this would impact clinical outcome. Further, the new-user study design attempts to mimic a trial study design [37] and excludes prevalent statin users who have tolerated statin treatment until study entry, which is essential as studies investigating prevalent and ongoing drug use has produced misleading findings [38,39]. Moreover, many breast cancer patients have existing comorbidities at diagnosis [40], which is associated with poorer survival [41]. Therefore, it would have been relevant to adjust for comorbidities in our models. Yet, this was not possible due to restricted data access.

In conclusion, this study investigated the association between post-diagnosis statin use and disease recurrence in a population-based cohort of patients with breast cancer. Post-diagnosis statin use was independently associated with a reduced risk of distant recurrence. Our study reinforces the need for a randomized clinical trial to equivocally determine whether statin treatment benefits patients with breast cancer.

Funding

This work was supported by , the , the , and the . The funding agencies played no role in the design of the study, the collection, analysis, or interpretation of the data, nor the preparation of the manuscript for publication.

Availability of data and materials

The patient data supporting the findings of this study are available to accredited academic parties upon reasonable request to the Malmö Diet and Cancer Study Steering Committee (https://malmo-kohorter.lu.se/malmo-cohorts).

Code availability

The code developed for this study is available upon reasonable request. Analyses were performed using Stata®, version 17 (StataCorp, College Station, TX, USA).

Authors' contributions

Study concepts (SH, SB), study design (SH, SB), statistical analyses (SH), interpretation of data (MI, MF, HJ, SH, SB), manuscript preparation (MI), manuscript editing (MI, SH), manuscript review (SB, HJ, MF, SH). All authors read and approved the final manuscript.

Ethics approval

The MDCS (LU 51–90) and the present study (Dnr 427/2007) were approved by the ethics committee in Lund, Sweden. All study participants provided their written informed consent upon enrolment.

Declaration of competing interest

The authors declare no competing interests.