The effect of non-oncology drugs on clinical and genomic risk in early luminal breast cancer

Abstract

Background: An effect of non-oncology medications on cancer outcome has been proposed. In this study, we aimed to systematically examine the impact of commonly prescribed non-oncology drugs on clinical risk and on the genomic risk [based on the Oncotype DX recurrence score (RS)] in early breast cancer (BC).

Experimental design: We collected data on clinical risk (stage and grade), genomic risk (Oncotype DX RS), and on non-oncology medications administered to 1423 patients with estrogen receptor-positive human epidermal growth factor receptor 2-negative BC during the month of their surgery. The influence of various medications on clinical and genomic risks was evaluated by statistical analysis.

Results: Out of the multiple drugs we examined, levothyroxine was significantly associated with a high Oncotype DX RS (mean 24.78; P < 0.0001) and metformin with a low Oncotype DX RS (mean 14.87; P < 0.01) compared with patients not receiving other non-oncology drugs (mean 18.7). By contrast, there were no differences in the clinical risk between patients receiving metformin, levothyroxine, or no other non-oncology drugs. Notably, there was no association between the consumption of levothyroxine and metformin and proliferation marker (Ki67) levels, but both drugs were significantly associated with progesterone-related features, suggesting that they influence genomic risk through estrogen-dependent signaling.

Conclusions: The results of this study indicate a significant impact of metformin and levothyroxine on clinical decisions in luminal BC, with potential impact on the clinical course of these patients.

Keywords: Oncotype DX; breast cancer; clinical risk; estrogen receptor; genomic risk; levothyroxine; metformin.

Figure 1

Non-oncology drugs are associated with different Oncotype DX recurrence scores (RSs). (A–C) Distribution and median expression levels of Oncotype DX RS. An unpaired Student’s t-test was used to test the significance for normally distributed data. The P values are indicated. (A) Metformin, levothyroxine, vitamin D, and selective serotonin reuptake inhibitor (SSRI) are associated with different median Oncotype DX RS compared with control (no known drugs) patients. (B) Differences in Oncotype DX RS for control (no known drugs) patients and metformin or levothyroxine groups according to patient age. (C) Differences in Oncotype DX RS between control (no known drugs) patients and metformin or levothyroxine groups according to histological subtypes of the tumors. ∗<0.05, ∗∗<0.01, ∗∗∗<0.001, ∗∗∗∗<0.0001. (D) Comparison of the frequency of patients presenting high or low clinical risk (CR) between control (no known drugs) patients and metformin or levothyroxine groups. BNZ, benzodiazepines; CCB, calcium channel blockers; IDC, invasive duct carcinoma; ILC, invasive lobular carcinoma; Inh, inhibitor; ns, not significant; PPI, proton pump inhibitors.

Figure 2

Levothyroxine and metformin influence genomic risk through estrogen-dependent modules. (A) Distribution and median expression levels of Ki67 (%) for control (no known drugs) patients and metformin or levothyroxine groups. (B–D) Differences in hormone receptor expression between control (no known drugs) patients and metformin or levothyroxine groups. (B) ER and PR mRNA expression. (C) ER and PR protein expression (immunohistochemistry). (D) HER2 mRNA expression. (E) A GSEA enrichment plot of the gene set ‘Kegg_Steroid_hormone_biosynthesis’. The preranked analysis was performed with fold-change-normalized gene expression data from 36 patients 2 weeks before and after metformin treatment. The fold change refers to normalized expression after treatment/normalized expression before treatment. (F) A connectivity map plot of the 20 perturbational classes most associated with the response of MCF-7 to metformin. (G) PR fold change as a result of metformin treatments and E2 presence. Histogram depicts the mean fold change (n = 3) relative to the mean of vehicle control ± SD. Unpaired two-tailed t-test was used to detect statistically significant differences between metformin treatment for the –E2 and +E2 groups. In the absence of E2, metformin treatment shows significant increase in PR expression (P < 0.0001), whereas in the presence of E2 no significant results were observed (P = 0.589). (H) Left: comparison between the frequency of patients presenting high or low genomic risk in MammaPrint between control (no known drugs) patients and metformin or levothyroxine groups. Right: distribution and median expression levels of the risk of recurrence as determined by the PAM50 test results for control (no known drugs) patients and metformin or levothyroxine groups. An unpaired Student’s t-test was used to test the significance for normally distributed data for all results. The P values are indicated. ∗∗<0.01, ∗∗∗<0.001, ∗∗∗∗<0.0001. ER, estrogen receptor; GSEA, gene set enrichment analysis; HER2, human epidermal growth factor receptor 2; IHC, immunohistochemistry; mRNA, messenger RNA; PR, progesterone receptor; SD, standard deviation.