Fluvastatin reduces proliferation and increases apoptosis in women with high grade breast cancer

Abstract

The purpose of this study is to determine the biologic impact of short-term lipophilic statin exposure on in situ and invasive breast cancer through paired tissue, blood and imaging-based biomarkers. A perioperative window trial of fluvastatin was conducted in women with a diagnosis of DCIS or stage 1 breast cancer. Patients were randomized to high dose (80 mg/day) or low dose (20 mg/day) fluvastatin for 3–6 weeks before surgery. Tissue (diagnostic core biopsy/final surgical specimen), blood, and magnetic resonance images were obtained before/after treatment. The primary endpoint was Ki-67 (proliferation) reduction. Secondary endpoints were change in cleaved caspase-3 (CC3, apoptosis), MRI tumor volume, and serum C-reactive protein (CRP, inflammation). Planned subgroup analyses compared disease grade, statin dose, and estrogen-receptor status. Forty of 45 patients who enrolled completed the protocol; 29 had paired Ki-67 primary endpoint data. Proliferation of high grade tumors decreased by a median of 7.2% (P = 0.008), which was statistically greater than the 0.3% decrease for low grade tumors. Paired data for CC3 showed tumor apoptosis increased in 38%, remained stable in 41%, and decreased in 21% of subjects. More high grade tumors had an increase in apoptosis (60 vs. 13%; P = 0.015). Serum CRP did not change, but cholesterol levels were significantly lower post statin exposure (P<0.001). Fluvastatin showed measurable biologic changes by reducing tumor proliferation and increasing apoptotic activity in high-grade, stage 0/1 breast cancer. Effects were only evident in high grade tumors. These results support further evaluation of statins as chemoprevention for ER-negative high grade breast cancers.

Introduction

Statins, by interrupting the biosynthetic pathway that produces mevalonate, affect several downstream pathways critical for cancer growth and progression, and through these mechanisms, have potential as anti-cancer agents. Many studies have examined the potential association between statin use and the risk of developing breast cancer, including ER-negative cancer, although the data from epidemiologic studies is not consistent [1–10] The meta-analyses of clinical trials designed with cardiovascular endpoints do not show an association between statin use and breast cancer risk [11–15]. Population-based studies have shown mixed results. However, the largest published cohort looking specifically at the impact of statin exposure on ER-negative tumors showed that the relative frequency of ER-negative breast cancers decreased in women taking lipophilic statins for more than a year (OR 0.63, 95% CI 0.43–0.92; P = 0.02) [16].

Three factors may contribute to these inconsistent findings. First, not all statins are alike. Some of the cardiovascular studies included in these analyses used both lipophilic and lipophobic statins and were later re-examined for cancer outcomes. Lipophilic statins diffuse across cell membranes and interrupt mevalonate synthesis both in the liver and in peripheral tissues, demonstrating the entire spectrum of cholesterol-dependent and independent effects. In contrast, lipophobic statins are taken up only via active transport, and hence affect hepatocytes that express the appropriate transporter molecule. When studies explicitly examine the impact of lipophilic versus lipophobic statins, some demonstrate a protective role of lipophilic statins in breast cancer prevention, [3, 6, 7, 10] but not all [2, 9, 14]. Second, the incidence of breast cancer in the cardiac risk patient populations is not high, which limits the power of these analyses to detect a difference between statin subtype. Third, and perhaps most important, the incidence of ER-negative disease among older women who would likely be taking statins is quite low. ER-negative cancers comprise less than 15% of breast cancers among the older postmenopausal U.S. and European population in statin studies designed to evaluate the reduction in cardiovascular morbidity and mortality.

We and others have shown that lipophilic statins inhibit growth of breast cancer tumor cell lines and breast cancer tumors in animal models [17–19]. These findings indicate that lipophilic statins may inhibit breast cancer cell growth and that impact is highly dependent on tumor characteristics. High grade, ER-negative cell lines (MDA-231) showed greater inhibition of tumor growth by lipophilic statins than did lower grade, ER-positive cell lines (MCF-7). Lipophilic statins, when administered in the drinking water, slowed tumor growth rates in an ER-negative mouse model [17]. Taken together, data suggest that ER-negative breast cancer is the biologic subtype most likely to be affected by lipophilic statins, an impact that epidemiologic studies could miss.

To obtain evidence of lipophilic statin effect and to define the target population in whom statins were likely to be effective, we designed a pilot window study to determine the biologic impact of short-term lipophilic statin exposure on in situ and invasive breast cancer. We hypothesized that proliferation would decrease more in high grade (grade 3) when compared to non-high grade (grade 1 or 2) tumors.

Patients and methods

Trial design and study subjects

We conducted a perioperative “window” trial in which a lipophilic statin would be prescribed during the interval between diagnosis by core biopsy and definitive surgical resection. The chosen treatment interval was 3–6 weeks (Fig. 1). Two doses of statin were tested.

Fig. 1.

Peri-operative window trial design

Women were eligible to participate if they had a histologically confirmed diagnosis of DCIS or stage 1 breast cancer. Subjects were accrued from women presenting at one of six breast care facilities in the United States: the University of California San Francisco and Marin General Hospital (CA), Dana Farber Cancer Center and South Shore Hospital (MA), University of Chicago (IL), and Memorial Sloan Kettering Cancer Center (NY). Exclusion criteria were neoadjuvant chemotherapy or hormonal therapy, current statin use, contraindication to statin therapy, or significantly elevated liver function enzymes at baseline (AST, ALT).

Participants (n = 40) were randomized 1:1 either to low dose (20 mg/day) or high dose (80 mg/day) fluvastatin. Fluvastatin was provided by Novartis Oncology (East Hanover, NJ). Paired biomarkers were performed before and after statin treatment. The study was approved by the institutional review boards of each site and informed consent was obtained from all participants.

Endpoints and biomarkers

The primary endpoint was change in the proliferative biomarker Ki-67 as determined by standard immunostaining of tumor tissue and light microscopic interpretation (Olympus, BX-41; Japan) by an academic surgical pathologist with expertise in breast pathology (FLB). Secondary endpoints were various tissue, serum, and imaging-based biomarkers (Table 1). A marker of tumor apoptosis, CC3, was investigated as an experimental endpoint. HER2 status was assessed by immunohistochemistry. Serum cholesterol served as the established indicator of statin effect and serum C-reactive protein, as an indicator of generalized inflammatory response. Paired magnetic resonance imaging was performed on a subsample of subjects and volume of enhancement was evaluated.

Table 1.

Biomarkers used as secondary endpoints to determine the biologic impact of short-term lipophilic statin exposure on in situ and invasive breast cancer

Biomarker	Indication	Measurement/Scoring	Source
Ki-67	Tumor proliferation	Percentage of 500 tumor cells staining positive	Tissue
CC-3	Apoptosis	Score	Tissue
		0 = 0 per HPF
		1 = 1%
		2 = >1–10%
		3 = >10–33%
		4 = >33–66%
		5 = >66%
Cholesterol	Fluvastatin effect	mg/dl	Serum
C-reactive protein	Generalized inflammation	mg/dl	Serum
Tumor volume	Tumor size	Cubic centimeters	Magnetic resonance imaging

Immunohistochemistry

Tissue immunohistochemical analysis was done before and after statin therapy on formalin-fixed, paraffin-embedded tissue sections cut at 4–5 μm using routine avidin–biotin immunoperoxidase technique (Vectastain Elite Kit, Vector Labs, Burlingame, CA). Antibodies and dilutions were Ki-67 (DAKO, Carpinteria, CA) at 1:100, rabbit anti-cleaved caspase-3 polyclonal antibody CC3 (Cell Signaling, Beverly MA) at 1:200, HER2 (Zymed Laboratories, South San Francisco, CA) at 1:200. Antigen was retrieved either by incubation with Ficin (Zymed) at 37°C (HER2), or by 0.01% trypsin digestion followed by heat treatment in 10 mM citrate buffer (Ki-67).

All tissue immunostained for Ki-67, CC3, and HER2 was reviewed by one breast pathologist blinded to randomization status, outcomes and without knowledge of other immunostaining results from the same specimen. Ki-67 proliferation was assessed, both in DCIS and invasive carcinoma, from the area with the highest mitotic activity, when possible, at least 1,000 tumor cells were counted in these areas and a Ki-67 proliferation index was obtained by calculating the percentage of Ki-67 positive tumor nuclei. CC3 staining was examined across the entire whole section and from areas of greatest CC3 staining, the percentage of CC3 positive cells was estimated on high power (400×) and a CC3 score category was assigned to each case: 0, 1, >1–10, ≥10–33, ≥33–66, ≥66%. HER2 was scored by standard ASCO/CAP guidelines [20]. The histologic grade of invasive carcinomas was assessed according to Nottingham criteria and the DCIS grading used nuclear grade [21]. Tumor grade was assessed and characterized as high grade (grade 3) or non-high grade (grade 1, 2).

Serum biomarkers

Serum total cholesterol and CRP levels were determined by the clinical laboratories of each institution according to standard clinical protocol from a 10 ml clotted blood sample.

MR imaging

MR Imaging was performed in a subset of patients using a contrast-enhanced T1-weighted fast gradient echo technique. High spatial resolution images were acquired before and following injection of gadolinium-based contrast. Following image acquisition, tumor volume was calculated using a previously described semi-automated software program [22]. Tumor volume was calculated as the sum of all pixels meeting a threshold value of 70%.

Statistical analysis

Within group comparisons (pre vs. post treatment) were made using paired t tests for normally distributed data (cholesterol and longest diameter) and the Wilcoxon signed-rank test for data that were not normally distributed (Ki-67). We chose the nonparametric Wilcoxon signed-rank test to compare pre and post Ki-67 index measurements because, although the distributions of Ki-67 index tended to be skewed and log-transformation of the pre and post treatment values led to approximate normality, the difference in log-transformed values failed the Shapiro–Wilk test for normality (P = 0.006). We used a Kruskal–Wallis test to determine whether median Ki-67 differences (pre minus post) differed by whether CC3 was unchanged, increased or decreased.

Between group comparisons were made using t tests for normally distributed data or the Wilcoxon rank sum test for data that did not follow a normal distribution. The statistical significance of dichotomous or continuous variables between treatment groups was determined using the chi square test or Fischer's exact test for groups with fewer than five observations. All statistical analyses were performed using Stata 10 (College Station, TX).

Results

A total of 45 patients consented and entered the study and 40 completed the study protocol. The sites and the number of patients who were accrued and completed the protocol included: University of California San Francisco and Marin General Hospital, 24/22; Dana Farber Cancer Center and South Shore Hospital, 17/15; University of Chicago, 3/3 and Memorial Sloan Kettering Cancer Center, 1/0. A total of 20 patients were randomized to receive the high dose of fluvastatin and 20 to receive the low dose.

Of the 40 subjects who completed the study protocol, 33 complete sets of tissue blocks were retrieved (83%). Twenty-nine were successfully immunostained, and four had insufficient tumor for immunostaining. Seven tissue blocks were not able to be retrieved: two missing, two were exhausted, two subjects did not have residual tumor at the time of surgery, and 1 patient opted not to undergo surgery.

The average age of patients at the time of enrollment was 55 years; 65% were ER-positive and 53% had DCIS (Table 2). One subject had a study related adverse event of elevated liver function tests (ALT, AST) after 80 mg/day of fluvastatin which resolved with discontinuation of the study medication.

Table 2.

Patient and tumor characteristics for enrolled subjects versus those with complete results for the primary endpoint, Ki-67

	Enrolled (n = 40)	Ki-67 data complete (n = 29)
Mean age (range)	55 (35–79)	56 (37–79)
ER positive	26 (65%)	18 (62%)
PR positive	25 (63%)	16 (55%)
HER2 positive	12 (32%)*	9 (31%)
DCIS	21 (53%)	16 (55%)
DCIS and invasive cancer	16 (40%)	10 (34%)
Invasive cancer only	3 (8%)	3 (10%)
Tumor grade
Grade 1	8 (20%)	5 (17%)
Grade 2	12 (30%)	9 (31%)
Grade 3	20 (50%)	15 (52%)

ER estrogen receptor, P progesterone-receptor, HER2 human epidermal growth factor receptor 2; DCIS ductal carcinoma in situ

^*n = 38

Primary endpoint: Ki-67

Overall, after statin treatment, the Ki-67 proliferation index was reduced in 20 tumors, increased in eight, and unchanged in one. The decrease in Ki-67 was significant in the 15 high grade tumors. Proliferation of high grade tumors decreased by a median of 7.2 percentage points (P = 0.008) with fluvastatin treatment over a treatment period that ranged from 21 to 50 days (median = 28 days). This decrease was statistically greater than was observed in low grade tumors (P = 0.04; see Fig. 2). Non-high grade tumors had a 0.3 percentage point decrease in tumor proliferation (P = 0.47). With respect to ER status, the change in Ki-67 index was not statistically significant (2.0% median reduction in ER-negative vs. 0.8% median reduction in ER positive, P = 0.56). Median reduction among the 20 HER2 negative subjects (−2.1%) and 9 HER2 positive subjects (0.2%), did not differ significantly (P = 0.67). There was no difference in effect between the 80 and 20 mg/day fluvastatin doses (P = 0.79) in terms of Ki-67 change pre and post statin exposure.

Fig. 2.

Change in the primary endpoint, Ki-67, by tumor grade

Secondary endpoints: CC3, serum cholesterol, and serum CRP values

Immunostaining for the secondary endpoint, CC3, a marker of apoptosis, was completed for the same 29 paired samples available for the Ki-67 staining. These results showed that tumor apoptosis increased in 38% of treated subjects, remained stable in 41%, and decreased in 21%. More high grade than non-high grade tumors had an increase in apoptosis (60 vs. 13%, respectively; P = 0.015; Fig. 3). A larger proportion of ER-negative tumors had an increase in tumor apoptosis versus ER-positive tumors (55 vs. 28%, respectively; P = 0.48), but this difference was not statistically significant. Tumors in which apoptosis decreased over the treatment period showed a greater median reduction in proliferation (−8.5 vs. −3% for those with increased apoptosis, and −0.2% for those whose apoptotic levels did not change with treatment; P = 0.08).

Fig. 3.

Change in CC-3, by tumor grade

Serum cholesterol was obtained before and after statin treatment for 35 subjects. It was significantly reduced after the treatment period (−34.9 mg/dl mean, 16%; P < 0.001) for all subjects. This reduction was greater in subjects who received high dose fluvastatin than in those who received the low dose (−44.2 vs. −26.1 mg/dl, P = 0.018) (Table 3). Paired serum CRP values were available for 29 subjects. The median change in serum CRP was 0% for both the high dose and low dose arms (Table 3).

Table 3.

Subgroup analysis showing change in biomarkers by tumor grade, estrogen-receptor status, and fluvastatin dose

Biomarker	n	Median Change	SD	Interquartile range	P value*	n	Median Change	SD	Interquartile range	P value*	P value**
		High grade (grade 3)					Low grade (grade 1, 2)
Ki-67	15	−7.2%	9.4	(−13.4, 0)	0.008	14	−0.3%	4.3	(−3.0, 0.8)	0.47	0.04
CRP	15	0%	40.7	(−42.9, 23.5)	0.75	15	0%	64.2	(−25, 9.1)	0.75	0.82
Cholesterol	18	−27.4	25.0	(−39.8, −15.0)	<0.001	17	−42.9	25.6	(−56.0, −29.7)	<0.001	0.96
		ER negative					ER positive
Ki-67	11	−2%	10.2	−13.4, 1.0	0.08	18	−0.8%	6.8	(−6.6, 0.8)	0.06	0.56
CRP	11	0%	45.9	(−47.5, 33.3)	0.88	18	0%	59.6	(−25.0, 12.5)	0.60	0.62
Cholesterol	12	−23.7	15.6	(−33.6, −13.7)	<0.001	22	−39.5	28.8	(−52.3, −26.8)	<0.001	0.04
		High dose fluvastatin (80 mg/day)					Low dose fluvastatin (20 mg/day)
Ki-67	14	−0.8%	10.2	(−7.2, 0)	0.06	15	−2.0%	6.2	(−9.8, 1.6)	0.06	0.79
CRP	13	0%	67.2	(−20, 0)	0.82	17	0%	39.3	(−41.2, 16.7)	0.63	0.66
Cholesterol Mg/dL	17	−44.2	30.2	(−59.8, −28.8)	<0.001	18	−26.1	18.2	(−35.1, −17.0)	<0.001	0.018

Mean and 95% CI are reported for the normally distributed variable of serum cholesterol

Median and Interquartile ranges are reported for variables that do not follow a normal distribution (Ki-67, CRP)

^*P value of group paired differences between pre and post measurement mean or median change from baseline (Wilcoxon signed rank test used for all P values except cholesterol measurements for which a paired t test was used)

^**P value of differences in high vs. low grade, ER status, and fluvastatin dose either mean or median change from baseline (Wilcoxon rank sum test used for all P values except cholesterol measurements for which a t test was used)

Bold values indicate statistically significant P values

MR imaging results

Volume analysis on 14 high-resolution paired MRIs showed that at baseline, median tumor volume was 2.68 cc (range, 0.13–17.6 cc), a 12.7% reduction in median tumor volume after treatment was observed which was not statistically significant (P = 0.27). Descriptively, nine tumors decreased in volume and five tumors remained stable or slightly increased in size. Seven of the nine lesions demonstrating some decrease in volume were high grade tumors, whereas four of the five tumors that were stable or slightly larger were low grade. Among the eight high grade tumors where there was paired imaging, volume decreased by a median of 24.8%, whereas in six paired lower grade tumors, volume increased by a median of 23% (P = 0.02).

Discussion

The goal of our pilot window biomarker modulation trial was to determine whether short-term exposure to a lipophilic statin, fluvastatin, had demonstrable anti-proliferative and pro-apoptotic effect on in situ and invasive breast cancer. The biology of statins, combined with our preclinical and epidemiologic investigations, led us to hypothesize that the biologic effect of statins would be most evident in high grade or ER-negative disease. Our findings support this hypothesis.

The cholesterol lowering activity of statins was evident within the brief window of the study, demonstrating that patients took the study agent and that it was active in the period of exposure. There was no evidence of CRP lowering during this period, suggesting that the window chosen may not have been sufficient to show a maximal anti inflammatory benefit.

Within the time limitations of the window trial design, we observed biologic activity on cholesterol, proliferation, and apoptosis. Effects were significant in high grade DCIS, not just ER-negative patients, although the small numbers of patients with paired samples may be a limiting factor. Seventy-three percent of the high grade lesions were ER-negative, however, four subjects with ER-positive high grade DCIS had significant changes in both Ki-67 and CC3, suggesting that other factors such as proliferation may determine sensitivity to statins. Moving forward, it would seem most appropriate to target high grade DCIS patients for a trial of statin impact, and to further stratify on ER status to better resolve the question of whether grade or ER status is the more appropriate criteria for patient selection.

Identifying a safe acceptable prevention agent for breast cancer continues to be challenging. Women often decide not to take the chemoprevention agent, Tamoxifen, to reduce breast cancer risk [23] despite evidence that it is relatively safe and reduces risk for ER-positive breast cancer [24]. Reasons for declining tamoxifen include increased (1) endometrial cancer risk; (2) thromboembolic events; and (3) vasomotor symptoms [24–26] Raloxifene has been shown to have a better safety profile [26], and it is prescribed and approved to reduce bone loss [27], so women have been more willing to accept this agent as a chemopreventive. What is most compelling to women in the prevention setting is knowledge that their risk for developing breast cancer is significantly elevated, that the preventative agent will reduce the risk of the cancer type they are most likely to develop, and that the medication risk profile is low.

Statins are appealing in that their safety and health promoting effects are well-established and they are well tolerated even at higher doses [28, 29]. Adverse event rates, even at high doses (atorvastatin 80 mg) are low, 0.6% with serious hepatic adverse events and 1.3% with musculoskeletal serious adverse events [30]. Statins reduce the chance of a cardiovascular event or stroke by 30% even in subjects with cardiovascular risk factors and normal cholesterol levels [31, 32]. In particular, the benefit of stroke risk reduction appears to be cholesterol independent [33–41]. Interestingly, Ki-67 reduction appeared to be independent of cholesterol in the window trial as well.

Ironically, the patient population commonly prescribed statins is older women who infrequently develop high grade/ER-negative breast cancer for which statins are most likely to be effective. A recent meta-analysis of cancer incidence in participants of the simvastatin and ezetimibe trials [12], showed that in one study of 1,873 patients, there were 166 total cancers, only 13 of which were breast cancer. In combined trials of 20,617 patients, only 40 of 639 cancers were breast cancer. The low number of cancer events in these trials limits what inferences may be drawn about the impact of statins on cancer risk. Our window trial provided evidence of biologic effect in women with high grade in situ or invasive breast cancer. There are two potential trial populations that are known to be at risk for this type of cancer. One is BRCA1 carriers, who are at risk for ER-negative high grade cancers preferentially. Statins are currently being evaluated in this population [42]. The other is women with high grade DCIS who choose lump-ectomy. The adjuvant setting of high grade DCIS could serve as a setting for evaluating the risk reducing potential of statins where recurrence events occur within the first 5 years.

Our study highlights the opportunities as well as the challenges of conducting window trials. The feasibility of obtaining paired endpoints in a multicenter setting was the primary limitation of this trial. It was particularly difficult to obtain pathology specimens from the initial cores if the biopsy was performed in an outside hospital. Scheduling paired MRI studies within a limited time window was challenging, and did not add sufficient value for the complexity it generated. Future studies with paired immunohistochemistry endpoints, particularly if one is a biopsy specimen, should be powered for greater dropout rates since endpoints will be lost in tissue processing. Finally, studies that include DCIS as an endpoint present a particular challenge because the amount of DCIS in a biopsy may be sparse. The development of universal fixatives that preserve histology, nucleic acids, and proteins would increase the number and types of analyses that can be performed on paired tissue samples, and likely eliminate or reduce the barrier to making diagnostic tissues available for research.

Conclusions

Our pilot study suggests that statins have biologic effects on the high grade subset of early stage breast cancers. Fluvastatin decreased tumor proliferation and increased apoptosis in these tumors, results that are concordant with the mechanism of action of statins and the results from preclinical experiments and some epidemiologic studies. Our findings also support further investigation of lipophilic statins as potential agents to prevent progression in high grade disease. Given its aggressive nature, greatest sensitivity to lipophilic statins, the lack of alternative interventions, and the safety and health promoting benefits of statins, we can ill afford to miss the opportunity to assess the effect of statins to prevent high grade breast cancer.