Ambient temperature is an important driver of human health, with well-established associations between ambient temperature changes and disease burden [17], mortality [18], and morbidity [19]. Preclinical and epidemiologic data has suggested an association between cancer progression and climate. Our analysis demonstrates that higher AAT was an independent predictor of improvement in OS, DSS, and pCR in stage I-III BC patients. We found that patients living in areas with an AAT ≥60.9°F had 10% higher odds of achieving a pCR. In the SEER cohort, patients living at AAT ≥ 47.5°F had a 16% improvement in OS and 15% improvement in DSS, whereas, in the NCDB cohort, a 5% improvement in OS was noted at optimal temperature cutoff of AAT ≥60.9°F. With every 5°F increment in AAT, a 4% improvement in OS and DSS was noted in the SEER cohort, whereas a 2% improvement in OS was noted in the NCDB cohort. The differences between SEER and NCBD cohorts may be attributed to the availability of more specific county-level data with the SEER cohort vs regional data with NCDB. The results also indicate that the impact of AAT on survival was more notable when data were grouped according to optimal temperature cut-off values. While the exact causes of such findings are unclear, we speculate that the greater impact at more extremes of temperature may indicate that the effect of temperature may not be linear and at lower or higher temperatures, a tipping point may be reached accounting for these differences.
The findings that a warmer climate is associated with favorable survival outcomes and a higher chance of achieving pCR in stage I-III BC patients is in line with our recent publication showing a trend towards worse survival in patients whose tumors have high thermogenesis score (which is shown to be increased in cold weather) [20]. We showed that high thermogenesis TNBC i.e., tumors under thermal stress, was associated with an immunosuppressive TME and a significantly worse DSS[20].
While the exact mechanisms and pathways underpinning the negative impact of cold climate on oncological outcomes have not been fully elucidated yet, we propose several explanations for these findings. We hypothesize that impaired anti-tumor immune responses in the TME under cold stress contribute to the observed worse outcomes amongst patients living in a colder climate. As previously mentioned, sub-thermoneutral temperatures were associated with impaired immune response in the TME. Housing mice under these temperatures resulted in a lower number of functional anti-tumor CD8+ T and overexpression of MDSCs [21], a potent suppressor of anti-tumor immune responses and CD8+ T proliferation [21,22]. Besides, Bucsek et al. found Tregs and programmed death receptor-1 (PD-1) overexpression in mice exposed to cold stress and showed that mice housed at thermoneutral temperature of 300C exhibited an anti-tumor immune microenvironment with higher number of CD8+ T cells and lower number of MDSCs and T-regs than mice that were kept at the sub-thermoneutral temperature of 220C [23].
pCR is a commonly utilized surrogate marker for long-term survival benefits in neoadjuvant BC clinical trials. Previous reports have demonstrated that pCR is significantly associated with prolonged event-free survival and OS, with a stronger association amongst patients with more aggressive BC subtypes (TNBC and HER2+) [24,25]. Thus, regulatory authorities have supported the use of pCR in BC trials as a surrogate endpoint for long-term survival outcomes for accelerated approval of therapies [26,27]. This study investigated the impact of pCR, alongside the traditional survival outcomes, owing to its well-established prognostic utility.
Our study findings that validate the association between temperature and outcomes in BC have clinical and translational implications. Our study validates the results of pre-clinical data noting an association between tumor behavior and temperature. It was also shown that housing temperature alters mouse biology and anti-tumor immune responses and sensitivity of tumors to therapeutic interventions [8]. Therefore, the housing temperature of cancer models should be considered during the assessment of therapeutic interventions, alongside the consideration of climate as a possible confounder during the clinical assessment of novel therapies. Building on this knowledge of immunosuppression in cold temperatures, targeted interventions for neural thermoreceptive pathways are proposed for improving clinical outcomes amongst patients living in a colder climate. In a phase I clinical trial, adding a non-selective β-blocker to pembrolizumab, an immune checkpoint inhibitor, led to an impressive objective response rate of 78% in patients with metastatic melanoma [28]. In BC models, non-selective β-blocker reversed the effects of cold stress on tumor growth and spread [29]. In a phase II clinical trial, 60 BC patients were randomized to receive propranolol or placebo one week preoperatively. The results showed that propranolol improved the anti-tumor immune response and reduced biomarkers associated with metastatic potential [30].
The drivers of tumorigenesis and alteration in the TME in those who live in colder environments can be analyzed in detail using metabolomics which involves the measurements of various small molecule metabolites including signaling mediators, nutrients, proteins in the blood, and the metabolic products of these molecules in the body fluids [31]. This would help us to develop personalized treatment options for those with aggressive malignancies who live in colder environments. As cold therapy for alopecia [32] and cryotherapy to prevent chemotherapy-induced neuropathy is gaining momentum [33], the clinical outcomes of patients who undergo these treatments are worth examining to scrutinize possible impact on their clinical outcomes with these novel interventions. In recent years, treatments involving hyperthermia have been utilized with success in multiple cancers. Hyperthermia when combined with chemotherapy and immunotherapy has shown promising results in gynecological malignancies treatments including Hyperthermic Intraperitoneal Chemotherapy (HIPEC) [34] and has increased therapeutic effectiveness in melanoma when combined with radiation therapy [35]. Given the finding from our study that shows that BC patients who live in warm climatic regions have a better prognosis, we could potentially utilize the benefit of hyperthermia treatments in BC in the future.
Our study has the strengths of a large sample size and representation of the US population. Also, we used two large national databases to investigate our hypothesis, and both showed similar findings. These findings were in line with our preclinical studies showing a significant association between ambient temperature and TME. Nonetheless, the study has several limitations. First, the analysis is based on retrospective data, which limits the control of the outcomes. BC subtype data were missing in a considerable proportion of the SEER cohort. Data regarding some socioeconomic variables, such as education and income, were not available in the SEER database, thus was not able to be adjusted for in this analysis. However, data on these variables was available in the NCDB analyses and was adjusted for. Despite these differences in the two databases, the clinical outcomes were similar. In terms of AAT, the NCDB provides regional location data; hence, we could not account for the variation in temperature across each region. Patients may have changed their residence during the course of surgery and follow-up, which can influence our results and therefore a time-dependent analysis might not be as informative. Since this is an ecological study examining the relationship between clinical outcome and temperature exposure at a population level, individual level data (e.g. air-conditioning, personal and lifestyle factors) was not available. Unlike mice, humans can mitigate cold stress by changing their environment. For example, some people are more exposed to the effects of ambient temperature, air pollution, light e.g., construction workers, vs. others who spend the majority of their time in a temperature-controlled setting, e.g., desk workers, and we were not able to account for this disparity in our analysis. The limitation of SEER was that it does not provide pCR data. Therefore, pCR analysis was done only with the NCDB. Another limitation of our study is that there is overlap among patients whose data is collected both, using SEER and NCDB databases. However, we decided to use both databases, because SEER provides us data on DFS while NCDB provides us data on pCR.