Fenretinide in Young Women at Genetic or Familial Risk of Breast Cancer: A Placebo-Controlled Biomarker Trial

Breast cancer, the most common cancer in women, is influenced by numerous risk factors such as age, obesity, unhealthy lifestyle, and genetic predisposition. Hereditary breast cancer accounts for 5% of all breast cancers and approximately 25% to 30% are related to BRCA1/2 pathogenic variants (1). Women with BRCA1/2 mutations have a high lifetime risk of developing breast and ovarian cancer (2, 3). Particularly, for BRCA1, pathogenic variants are associated with breast tumors characterized by early onset, and worse prognosis (4). Thus, there is an urgent need to enhance methods of screening and improve prevention medicine by investigating new preventive agents (5). Retinoids are natural or synthetic vitamin A derivatives with a role in the regulation of cell growth, differentiation, and apoptosis (6). They have been extensively evaluated in clinical trials due to their anticancer potential (7).

Retinoids are still being investigated for breast cancer prevention, as documented by a large number of preclinical studies and clinical trials (6). In particular, fenretinide (N-4-hydroxyphenyl-retinamide), a synthetic derivative of all-trans retinoic acid, is the most studied retinoid in breast cancer chemoprevention trials due to its selective accumulation in breast tissue and its favorable tolerability (8) and safety profile (9). Fenretinide exerts its activity through the binding to nuclear receptors leading to the regulation of several cellular processes, including cell differentiation and apoptosis (7, 10). In addition, fenretinide inhibitory effects seem to be partly independent of the hormonal receptors’ status, because it can induce apoptosis in both ER-positive and ER-negative breast cancer cell lines (11). This agent showed a significant reduction in the incidence of second breast tumors in premenopausal women confirmed after a 15-years follow-up. Interestingly, in the same trial, the incidence of ovarian cancer during the 5 years intervention period was significantly lower in the treatment arm (12). Indeed, fenretinide is highly effective in inhibiting the growth of BRCA1 mutated breast cancer cell lines (13).

Given these encouraging findings, we designed a double-blind placebo-controlled randomized phase III trial to investigate fenretinide efficacy in reducing the incidence of breast cancer in healthy young premenopausal women at increased familial/genetic risk for breast cancer, or women with previous breast cancer who have concluded their adjuvant treatment. Because of the very low rate of patient accrual, the trial was closed early. Thus, here we present only the circulating biomarker results of fenretinide versus placebo. Specifically, we measured circulating biomarkers associated with metabolic health and breast cancer risk such as insulin and blood glucose, the insulin growth factor (IGF) system, androgens, and retinol binding protein-4 (RBP-4), at baseline, after 12 and 36 months of treatment.

The study was conducted at four Italian centers: European Institute of Oncology in Milan, Centro di Riferimento Oncologico in Aviano, Azienda Ospedaliera Policlinico in Modena, and Ospedale S. Bortolo in Vicenza.

Eligible subjects were women 20 to 46 years old with a known BRCA1/2 mutation or with a ≥20% risk of being a mutation carrier according to the BRCAPRO model (14); with or without previous breast cancer. Any standard adjuvant treatment had to be completed. No other previous malignancies, excluding cervical intraepithelial neoplasia and nonmelanoma skin cancer, were allowed. All participants had to avoid pregnancy during treatment and for 12 months after drug cessation.

Participants were randomized to receive fenretinide 200 mg/day (100 mg soft gelatine capsules/twice daily) or a placebo for five years. A monthly 3-day interruption was introduced to allow the partial recovery of retinol levels, thus allowing sufficient uptake for correct night vision.

Solmag SpA manufactured fenretinide, whereas the formulation of the capsules and matching placebo were provided by the IEO pharmacy. The study was approved by the local Institutional Ethical Committees (EudraCT number-OsSC: 2009–010260–41) and was conducted in accordance with Declaration of Helsinki. All participants signed written informed consent. Women were stratified according to disease status (breast neoplasms: yes/no), and center.

The study was terminated in December 2017, because of the low accrual rate. Compliance was evaluated by a patient self-report, a calendar packaging (each subject received a 6-month calendar as a reminder of drug intake), and by measuring drug and metabolite concentrations in plasma. Adverse events were assessed using the National Cancer Institute Common Toxicity Criteria (version 3).

Fasting blood samples were collected at baseline, 12 and 36 months. Plasma heparin aliquots for drug, metabolite, and retinol measurement, as well as serum aliquots for other biomarkers, were stored at −80°C until assayed. Lab personnel were blinded to the allocated arm.

Serum glucose, total cholesterol (TC), high-density cholesterol (HDL), and triglycerides (TG) were measured on fresh samples according to routine procedures at the local laboratories. Low-density cholesterol (LDL) was calculated from the Friedewald formula: LDL = TC–(HDL+TG/5) (15), expressed in mg/dL. All other biomarkers were determined on frozen samples. ELISA were used for the measurements of serum RBP-4 (Immunodiagnostik AG), IGF-binding protein-3 (IGFBP-3; R&D Systems), vascular endothelial growth factor (VEGF) (R&D Systems), leptin (R&D Systems), and adiponectin (R&D Systems) following the manufacturer's instructions. The limit of detection of the tests was 0.9 μg/L for RBP-4, 0.05 ng/mL for IGFBP-3, 9.0 pg/mL for VEGF, 7.8 pg/mL for leptin, and 0.246 ng/mL for adiponectin. During the assay runs, in addition to the specific controls provided, an in-house pooled control sample was used to monitor the coefficient of variation between assays. The interassay coefficient of variation (run in duplicate at each assay run) was 10% (mean: 33.5 mg/L, n = 18) for RBP-4, 2.6% (mean: 409 pg/mL, n = 8) for VEGF, 6.0% (mean: 12.6 μg/mL, n = 18) for adiponectin, and 18.3% (mean: 14.6 ng/mL, n = 24) for leptin.

Serum concentrations of insulin and sex hormone binding globulin (SHBG) were measured by chemiluminescent microparticle immunoassays designed for the ARCHITECT analyzer (Abbott Laboratories).

We applied the homeostasis model assessment (HOMA) as a surrogate index of insulin resistance, obtained by the formula [fasting insulin (mU/L) × glucose (mmol/L)]/22.5 (16).

Serum IGF-I and testosterone were determined by chemiluminescent immunometric assays (Nichols Institute Diagnostics) performed on LIAISON autoanalyzer (Diasorin). The sensitivity of the assay was 5 pg/mL, and intra- and interassay coefficients of variation were 3.9% and 6.3%, respectively.

Per protocol analysis we measured plasma concentrations of fenretinide, its main metabolite N-(4-methoxyphenyl) retinamide (4-MPR), 4-oxo-N-(4-hydroxyphenyl) retinamide (4-oxo-4-HPR), and retinol. Plasma heparin samples were assayed by high-performance liquid chromatography, as previously described (17, 18). Briefly, 200 μL of plasma was combined with 400 μL of CH3CN containing 125 μg/mL butylated hydroxytoluene (BHT; Sigma). After vortexing and centrifugation to separate proteins, the resulting supernatants were analyzed using an LC equipped with a C18 reverse-phase column (5-μmol/L) and a C18 pre-column. The mobile phase used consisted of CH3CN:H2O:CH3COOH (75:23:2, v/v/v) at a flow rate of 2 mL/min. Detection was performed using a Perkin-Elmer LC95 absorbance detector at 370 nm.

Baseline biomarker levels and their absolute changes after 12 and 36 months were expressed using median and IQR. Medians comparison between treatment groups was assessed using nonparametric Wilcoxon test. Modulation of the HOMA score and HDL cholesterol from baseline to 12 and 36 months was also represented using box and whiskers plots. Analyses were performed with the SAS software version 9.4. All P values are two-sided and those <0.05 are considered statistically significant.

The data generated in this study are not publicly available because they could compromise patient privacy or consent, but are available upon reasonable request from the corresponding author.

A total of 265 premenopausal women were contacted to participate in the clinical study as shown in the consort diagram (Fig. 1). Of 133 screened women, 62 premenopausal women with a median age of 36 years (20–46) were randomized to receive fenretinide (200 mg/day) or placebo. The main characteristics of participants are shown in Table 1.

Overall, 53 participants were BRCA mutation carriers, of which 9 had a previous diagnosis of breast cancer. Only 9 participants were included because of an increased risk of being a BRCA mutation carrier (≥20% based on the BRCAPRO model). Participants were well balanced between arms for baseline characteristics.

Compliance was high, as only 2 (6.9%) of 29 patients randomized to receive fenretinide with serology performed at 12 months resulted in having no (1 patient) or very low (1 patient) concentrations of fenretinide or its metabolites. Median [IQR] of drug and metabolite concentrations at 12 months were for 4-HPR 134 [80, 195] g/mL, 4–0×0–4-HPR 80 [40, 112] ng/mL, and 4-MPR 127 [100, 157] ng/mL.

The most expected fenretinide side effects are visual and skin disorders. Participants in the fenretinide group had very mild visual side effects. In particular, five subjects in the fenretinide arm reported mild-to-moderate diminished dark adaption and vision, and three subjects in the placebo group claimed mild diminished dark adaption and vision.

No differences between arms were observed in the mean circulating biomarker levels at baseline (Table 2).

Table 3 summarizes the absolute changes in serum biomarkers from baseline levels. After 1 year of treatment, we observed a favorable effect of fenretinide on serum glucose, insulin, and HOMA index. Glucose absolute change (mg/dL) was −4.50 in the fenretinide arm, compared with 1.00 in the placebo group (P = 0.005). Insulin absolute change (uU/mL) was −0.55, compared with 0.50 (P = 0.03), and HOMA index was −0.28, compared with 0.22 in the placebo arm (P = 0.004). HDL cholesterol showed a significant increase (P = 0.002) in the fenretinide arm of 9.0 mg/dL, compared with placebo (−6.0 mg/dL). After 36 months, there was no significant difference between arms considering the absolute change from baseline, except for the marked reduction of retinol (P < 0.0001), and RBP-4 (P < 0.0001).

Figure 2 is a Box Plot presentation for HOMA index and HDL cholesterol to show more specifically the modulation from baseline to 12 and 36 months. Panel A shows the median level of HOMA index, which decreases in the fenretinide arm from 1.32 to 1.17 at 12 months (P = 0.01) and to 1.25 at 36 months (P = 0.71), whereas it increases in the placebo group from 1.09 to 1.45 at 12 months (P = 0.07) and to 1.22 at 36 months (P = 0.60). The opposite trend is shown for HDL cholesterol (Panel B), with increasing levels in the fenretinide arm and decreasing levels in the placebo.

Our present trial focuses on biomarkers modulation by fenretinide in a prevention clinical trial in young women at increased familial breast cancer risk, the population who should benefit from fenretinide treatment. The potential clinical benefit of fenretinide as a preventive agent was previously shown by Veronesi and colleagues in a phase III trial in women with early breast cancer. Overall, no significant benefit of second breast cancer was detected, but in a post hoc analysis, fenretinide showed in premenopausal women a borderline significant reduction for the contralateral events and a significant hazard ratio of 0.65 [95% confidence interval (CI), 0.46–0.92] for ipsilateral events (12). The 15-year follow-up reinforced this observation showing for overall, ipsilateral, and contralateral second breast cancer a significantly reduced event rate in the fenretinide arm (19).

The main findings of the presented study were the favorable effects of fenretinide on biomarkers related to glycemic and lipid metabolisms.

Specifically, we observed remarkable decreases in glucose, insulin, and HOMA index, whereas, for the lipid profile fenretinide increased HDL cholesterol, compared with placebo. The effects were seen at 12 months whereas at 36 months were no more evident. This observation was not due to lower compliance as it can be seen by the plasma levels of fenretinde and its metabolites. The small sample size was even smaller at 36 months and the data have to be taken as exploratory anyway, but we cannot rule out a possible drug resistance.

As expected, retinol and RBP-4 were greatly reduced by the active treatment and maintained stable over time. No effects on the IGF system were found, contrary to previous studies showing a moderate IGF reduction by fenretinide compared with controls (20, 21).

Metabolic syndrome, a condition that includes visceral adiposity, hyperglycemia, hypertriglyceridemia, and hypertension, is associated with an increased risk of breast cancer and a poor breast cancer prognosis (22).

There is biological evidence that glucose and other factors related to glucose metabolism may contribute to breast cancer development (23). Glucose may play a direct role in the development of breast cancer by favoring the growth of potentially malignant cells which require glucose for progression (23). In particular, its conversion to lactate is an important stimulus for cancer cell growth (24). Some studies have suggested an increased risk of breast cancer associated with elevated fasting serum glucose in nondiabetic subjects (25). This connection is thought to be linked to factors like insulin resistance, which can lead to higher levels of insulin and insulin-like growth factors (26). Glucose excess induces hyperinsulinemia, and also insulin has a direct or through the IGF system proliferative effect on breast cancer cells that may synergize sex hormones and adipokines (27). Somehow this process may become self-sustaining because prolonged hyperglycemia induces insulin resistance leading to inefficient glucose utilization with consequent hyperglycemia (28). Fenretinide's impact on insulin resistance has also been linked to its ability to inhibit dihydroceramide desaturase (DES1), an enzyme involved in de novo biosynthesis of ceramides, which counteracts insulin's effects. Blocking DES1 via fenretinide has been demonstrated to prevent lipid-induced insulin resistance by reducing cellular ceramides while increasing precursor dihydroceramides (29).

In a previous randomized clinical trial, we found that 2 years of fenretinide treatment in overweight premenopausal women improved insulin sensitivity and decreased serum leptin levels (30). High-fasting insulin levels can be considered a breast cancer risk factor (31) even for premenopausal women (32), with a stronger association for estrogen receptor-negative breast cancer (33). However, in a prospective study, circulating insulin levels were not associated with breast cancer incidence (34). Taking singularly, each biomarker of insulin resistance showed a weak association with breast cancer risk (35).

We found also a favorable effect of fenretinide on HDL that significantly increase after 12 months, whereas TG and LDL remained stable. This effect is confirmed by previous studies both in patients with metastatic breast cancer (36, 37) and healthy postmenopausal women (9, 38). HDL cholesterol has multiple functions beyond its role in cholesterol transport, exhibiting antioxidant and anti-inflammatory effects, protecting blood vessels from oxidative stress and inflammation, it can have either pro or anti-apoptotic activities (39). Low HDL cholesterol has been reported as a potential breast cancer risk biomarker because low HDL cholesterol is associated with high-free estradiol and high LDL/HDL with a higher BMI (40). Furthermore, low HDL was also associated with higher levels of androgens (41), and other steroids involved in breast cancer risk (42–45). However, previous studies on the association of HDL and estradiol, and cancer risk have reported contradictory results (46–48).

In the Italian phase III trial, we observed that fenretinide induced a moderate decline in IGF-I only in women ≤50 years of age, which only marginally explained the observed cancer risk reduction in women <50 years of age (21, 49). An explanation for the fenretinide associated reduction of IGF-I levels in premenopausal women arises from detailed studies explaining the cross-talk between estrogens and retinoids pathways in the regulation of IGF-1 gene (50, 51).

It is important to note that several genetic and non-genetic factors can influence circulating levels of IGF-I (52, 53) and can contribute to variations in the response to fenretinide treatment. Also, the timing of IGF-I level measurements during fenretinide treatment may impact the observed results (54) considering that IGF-1 levels can change during different phases of the menstrual cycle, and this fluctuation could be influenced by the changing levels of hormones, such as estrogen and progesterone (55).

Fenretinide has been shown to modulate retinol plasma levels thanks to a high binding affinity for RBP-4, which results in competition with retinol for the RBP binding site. In the liver, it binds RBP-4 reducing its secretion (56, 57). Furthermore, fenretinide promotes the renal clearance of RBP-4 by preventing the interaction between RBP-4 and transthyretin leading to elevated levels of RBP-4 in the urine and lowering of circulating levels of RBP-4 and retinol (57, 58). Both mechanisms synergize in reducing circulating serum levels of RBP-4 and retinol (59). Retinol metabolism is known to play an important role in metabolic disorders such as obesity, diabetes, and dyslipidemia (60, 61). In 2005, Yang and colleagues reported a study linking RBP-4 to insulin resistance (62). In normal mouse models, overexpression of human RBP-4 elicits insulin resistance and reduces glucose tolerance, whereas the genetic deletion of RBP-4 enhances insulin sensitivity (62). It was observed that mice fed with a high-fat diet and treated with fenretinide normalized their circulating RBP-4 levels as well as insulin sensitivity and glucose tolerance were improved (62). A strong direct correlation between RBP-4 and insulin resistance was demonstrated by Graham and colleagues (63) in humans. The underlying mechanisms linking fenretinide RBP-4 reduction with favorable effects on insulin resistance and obesity are still unclear. A putative association between fenretinide and improved metabolic profile can probably be attributed to the reduction in serum RBP-4 (62, 64). Nevertheless, the anti-obesity action of fenretinide can be also independent by the RBP-4 lowering and has been referred to the modulation of retinoid homeostasis genes (65).

A limitation of our study was the small sample size and the results had to be taken as exploratory. The lack of effects at 36 months could be for the further sample size reduction but we cannot exclude the development of drug resistance after a longer treatment, even though the effects on retinol and RBP-4 may not support this hypothesis. Another point is that we did not synchronize the blood draws according to the ovarian cycle. As IGF-I levels are associated with estradiol levels, this may have increased intrasubject variability and thus made it more difficult to observe an effect in such a small group. After 12 months of fenretinide, a decrease in IGF was observed compared to placebo, but it did not reach statistical significance.

In conclusion, our study has shown that fenretinide has a beneficial influence on the metabolic profile. These effects together with the other biological processes, including anti-angiogenic and anti-invasive pathways, may contribute to breast cancer prevention activity (66). Moreover, fenretinide, similarly to IRX4204, a novel antagonist of the nuclear retinoid X receptors, may have specific effects for BRCA1/2 mutated carriers (67). Given the complex nature of carcinogenesis, validated surrogate biomarkers are crucial in preventive medicine. Relying solely on a single biomarker such as insulin, IGF-I, or HOMA index might not sufficiently confirm the effectiveness of a preventive drug. Therefore, combining multiple relevant biomarkers and employing various endpoints and algorithms is recommended to better ascertain the role of a preventive drug. Taken together, the activity in glycemic and lipid metabolisms by fenretinide could have some impact in preventing the carcinogenic process.

L. Cortesi reports grants and personal fees from Astra Zeneca and MSD; personal fees from Pfizer, Novartis, Gilead, and Roche outside the submitted work. No disclosures were reported by the other authors.

V. Aristarco: Data curation, investigation, visualization, writing–original draft. D. Serrano: Supervision, investigation, methodology, writing–original draft, writing–review and editing. P. Maisonneuve: Resources, data curation, formal analysis, visualization. A. Guerrieri-Gonzaga: Data curation, funding acquisition. M. Lazzeroni: Resources, investigation. I. Feroce: Resources. D. Macis: Resources, data curation, investigation, methodology, writing–original draft. E. Cavadini: Resources, data curation. E. Albertazzi: Data curation. C. Jemos: Resources. E. Omodeo Salè: Resources. L. Cortesi: Resources, data curation. S. Massarut: Resources, data curation. M. Gulisano: Resources, data curation. M.G. Daidone: Resources, investigation, methodology. H. Johansson: Conceptualization, resources, data curation, supervision, investigation, methodology, writing–original draft, project administration, writing–review and editing. B. Bonanni: Conceptualization, funding acquisition, project administration, writing–review and editing.

A special thanks to Prof Umberto Veronesi, who was the main Italian investigator in cancer prevention medicine. This work was partially supported by: Italian Ministry of Health through the Integrated Program in Oncology (PIO 2006, PPS-2006–339673 - La prevenzione del cancro: sviluppo di modelli di intervento basati sull'evidenza; Lead: Regione Toscana); Ricerca Corrente 2007 through Tevere project (Breast cancer chemoprevention with Fenretinide in young women at increased risk); Fondazione Italiana Ricerca sul Cancro (FIRC).