The insulin sensitivity Mcauley index (MCAi) is associated with 40-year cancer mortality in a cohort of men and women free of diabetes at baseline

Background The association between insulin resistance and cancer-mortality is not fully explored. We investigated the association between several insulin sensitivity indices (ISIs) and cancer-mortality over 3.5 decades in a cohort of adult men and women. We hypothesized that higher insulin resistance will be associated with greater cancer-mortality risk. Methods A cohort of 1,612 men and women free of diabetes during baseline were followed since 1979 through 2016 according to level of insulin resistance (IR) for cause specific mortality, as part of the Israel study on Glucose Intolerance, Obesity and Hypertension (GOH). IR was defined according to the Mcauley index (MCAi), calculated by fasting insulin and triglycerides, the Homeostatic Model Assessment (HOMA), the Matsuda Insulin Sensitivity Index (MISI), and the Quantitative Insulin Sensitivity Check Index (QUICKI), calculated by plasma glucose and insulin. Results Mean age at baseline was 51.5 ± 8.0 years, 804 (49.9%) were males and 871 (54.0%) had prediabetes. Mean follow-up was 36.7±0.2 years and 47,191 person years were accrued. Cox proportional hazard model and competing risks analysis adjusted for age, sex, country of origin, BMI, blood pressure, total cholesterol, smoking and glycemic status, revealed an increased risk for cancer-mortality, HR = 1.5 (95% CI: 1.1–2.0, p = 0.005) for the MCAi Q1 compared with Q2-4. No statistically significant associations were observed between the other ISIs and cancer-mortality. Conclusion The MCAi was independently associated with an increased risk for cancer-mortality in adult men and women free of diabetes and should be further studied as an early biomarker for cancer risk.


Introduction
Cancer remains one of the most common causes for morbidity and mortality in the US and worldwide (1,2). Worldwide, 9.5 million cancer-related deaths were reported in 2018 (3).
In the US, cancer incidence is approximately 442.4 per 100,000 per year (2) and although declining, cancer death rate is 158.3 per 100,000 per year, placing it as the second most common cause of death in the US (1). According to predictions, 606,520 people died from malignancy in the US during 2020. In Israel cancer is also a leading cause of death with 13,050 deaths (25.4% of all cases of death, cumulative risk 9.09) reported in 2018 (4). A number of factors were associated with cancer incidence and carcinogenesis such as smoking (5), Body Mass Index (6), diabetes (7) and sedentary lifestyle (8).
The association between insulin resistance (IR) and cancer remain uncertain. Metabolic alterations were previously found to correlate with both IR and cancer through dietary risk factors (e.g. hypercaloric diet, low bers etc.) that induces in ammation and oxidative stress, or promote IGF-1 secretion that acts as a strong mitogen (9). The common soil hypothesis suggest that in susceptible individuals, metabolic Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js abnormalities such as obesity, IR and dyslipidemia would be the initial manifestation of unhealthy diets and lifestyle, whereas carcinogenesis is more prolonged with delayed clinical manifestations (10). A wide epidemiological evidence is showing that diabetes is strongly associated with speci c types of cancer (6), mainly pancreatic and liver cancer (11) and the American Diabetes Association and the American Cancer Society issued a consensus report on diabetes association with cancer incidence (12). Nevertheless, the nature of this association is yet to be clari ed, and there is a possibility of an indirect association, underlined by the hyperinsulinemic state or by glucose lowering medication use in addition to common risk factors such as obesity (13)(14)(15).
The associations between type 2 diabetes, IR and increased fasting glucose plasma levels with malignancy associated mortality were demonstrated in a number of studies (15,16), however these studies were mostly on diabetic participants or with a short follow up period. Furthermore other studies on diabetic participants including meta-analyses did not show such an association with cancer mortality (17).
This study aimed to investigate the association between insulin resistance surrogates, i.e. fasting insulin and glucose plasma levels and ISIs, with cancer mortality in an Israeli cohort study-the Glucose Intolerance, Obesity and Hypertension (GOH) study over a 40-year follow-up.

Study design and population:
Study population was previously described (26). The cohort was part of the second stage of the Glucose Intolerance, Obesity and Hypertension (GOH) study (27). The study is an ongoing prospective longitudinal study, initially began on 1967 and included a total of 8400 Israeli Jews, strati ed according to sex, ethnicity as determined by country of birth or that of the mother for Israeli born (Yemenite, Asian, North Africans, and European-North Americans) as well as birth decade (1912-1921; 1922-1931; 1932-1941).
The second phase was performed between 1979 and 1982 and included anthropometric measurements, blood tests and in particular fasting glucose and insulin plasma levels taken during oral glucose tolerance test (OGTT).
Participants with both fasting glucose and insulin plasma levels and free of diabetes were included in the present analysis. The nal sample included 1612 participants who met the inclusion criteria out of 2769 Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js participants primarily examined in the second phase. Information regarding the GOH population and methodology is further detailed elsewhere (24,27).
Blood glucose was measured using the automated Technicon Autoanalyser II (Technicon Instruments Corp, Tarrytown, NY); Blood insulin was measured using the Phadebas Radioimmunoassay kit (Pharmacia Diagnostics Inc. Piscataway, NJ). Blood test analysis was performed by a single laboratory.
Participants were followed until December 2016 for malignancy associated mortality. Participant's approval was obtained a priori and the study protocol was approved by the Sheba Medical Center's IRB.

Insulin resistance:
The current study examined IR state as re ected by the following IR surrogates and ISIs: Fasting insulin and glucose plasma levels: both were categorized into quartiles and the upper quartile (Q 4 ) was compared to lower quartiles (Q 1−3 ) as with the ISIs.
ISIs were calculated as follow (18-21, 28): Homeostatic model assessment (HOMA)-Insulin resistance (IR) and beta cell function (%B) (21), were calculated as follows: MISI mean glucose was calculated using glucose taken at 0, 60 and 120 minutes after an ingestion of 100 gr of glucose during an OGTT. Mean insulin was calculated using insulin taken at 0, 30, 60 and 120 minutes during the OGTT. level of signi cance in order to evaluate differences between those who remained alive, those who died from cancer and those who died from other causes by the end of the follow-up. The association between ISIs and 40-year malignancy associated mortality rate were examined for cumulative incidence analysis using the Cox proportional hazards model. Study participants were censored at the time of non-cancer deaths or by the end of follow-up, whichever came rst. An additional approach used death from noncancer causes as a competing risk to cancer death (the Fine and Gray method (29)) by calculating the sub-distribution hazard ratio (SHR). This method is based on the Cumulative Incidence Function (CIF) that counts failures from competing events and deaths from the primary endpoint, whereas the competing events in the cumulative incidence method are censored. Each insulin resistance surrogate was evaluated in a separate model. In order to avoid multicollinearity, Spearman's rank correlation coe cient test was performed, excluding covariates with a correlation of 60% or above from the same model. Survival analysis was performed according to cause speci c mortality (i.e. deaths from cancer vs survival and non-cancer deaths). Models were adjusted for demographic variables as for known mortality risk factors such as smoking status, blood pressure, BMI, cholesterol and diabetes status. Models are presented with Hazard Ratio (HR) or SHR with 95% con dence intervals (95%CI). The proportional hazards assumption was tested using the log minus log plot and by constructing an interaction variable composed of time-to-event multiplied by the covariate and entering it into the model.
Kaplan Meier survival curves for insulin resistance surrogates were compared using the log-rank test.
Statistical analysis was performed using SPSS version 25.0.

Baseline characteristics:
A total of 1612 subjects were followed until December 2016 for malignancy associated mortality. As presented in Table 1, only origin did not showed a signi cant difference between vital status groups. Prediabetes was found in 148 (56.1%), 427 (60.5%) and 296 (46.1%) individuals who died from cancer, from other causes, and survivors respectively. Compared to survivors, individuals who died from cancer were predominantly males, smokers and pre-diabetic, with increased blood pressure and BMI as well as with an increased fasting glucose, fasting insulin and total triglycerides plasma levels. Individuals who died were more frequently found in the IR quartiles of ln MISI, ln HOMA-IR, ln HOMA-%B, QUICKI and MCAi.
Individuals who died from other, non-cancer related, primarily cardiovascular causes, were older (P<0.001), with increased systolic (P=0.03) and diastolic (P=0.02) blood pressure, as well as higher total cholesterol (P=0.03) compared with individuals who died from cancer (not shown).  Adjusted survival curves using Cox regression ( gure 1) showed a signi cant shorter times until cancer death for individuals in the MCAi quartile (Q 1 ), p=0.004 ( gure 1).
An interaction between MCAi and glycemic state was not found to be statistically signi cant (pinteraction = 0.1).
Separate analyses were conducted according to glycemic status (i.e. normoglycemia and prediabetes). In the pre-diabetics group (n=850), both cumulative incidence analysis using the Cox proportional hazards model and the competing risk analysis demonstrated an increased risk for cancer mortality for MCAi Q 1 , HR=1.6 (95% CI: 1.

Discussion
In this long-term follow up of 1,612 men and women free of diabetes, a signi cant association was demonstrated between insulin resistance, as measured by the MCAi, and cancer mortality. No such association was found for the other IR surrogates.
These results reinforce the contribution of IR on the pathophysiology of cancer, exempli ed by the 40-50% increased risk for cancer related mortality and speci cally in individuals with prediabetes.
A number of previous studies have reported an association between increased fasting glucose plasma levels and cancer mortality (14)(15)(16)30 Such a positive association with increased cancer mortality was not observed (37)(38)(39) and even correlated with better disease-free survival (40) in breast cancer patients. Further investigation is needed in order to establish triglycerides inter-relationship with cancer progression and prognosis.
In the GOH cohort, an increased risk for all-cause mortality was found in individuals in the IR quartiles of the MCAi, the QUICKI and the HOMA-IR (26). However, the MCAi was the only ISI that showed a signi cant association with cardiovascular mortality, regardless of the presence of diabetes. The current ndings suggest that the MCAi may be used as a surrogate for the long-term increased risk of death for both malignancy and cardiovascular morbidities.
The current study did not include participants with the diagnosis of diabetes due to the potential confounding effect of anti-diabetes medications (39,41), as well as the established association between diabetes and cancer mortality (15,16). For example, medications for the treatment of diabetes such as Metformin were negatively associated with mortality among diabetic patient (39) while exogenous insulin use and Sulfonylureas were associated with an increased risk for cancer mortality (41). However these ndings are controversial, due to potential methodological aws (42). In addition, diabetic patients display distinct characteristics such as relatively low levels of endogenous insulin as part of the disease progression, and higher BMI, which may confound the association. As previously mentioned, an association between hyperinsulinemia and increased risk for cancer death was observed in a number of studies and thus, lower levels of insulin could potentially have a protective effect from cancer mortality (32). Moreover, studies have shown a negative association between the metabolic syndrome, and speci cally obesity, with cancer risk, and better outcome in cancer patient, suggesting that the metabolic syndrome and increased BMI may serve as good prognostic markers (43).
Strengths and weaknesses of the study Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js While the standard oral glucose tolerance test (OGTT), as recommended by the American Diabetes Association (44), require the oral administration of 75 gr glucose, in the current study the test was carried out using 100 gr of glucose. This was done due to the absence of clear guidelines at the time of the examination (1979)(1980)(1981)(1982). Furthermore, the ingestion of 100gr of glucose has been shown to improve insulin sensitivity and insulin secretion with minimal effect on the results of the OGTT in terms of the plasma glucose levels measured throughout the test (45).
In addition, the Yemenite population was over sampled in the GOH cohort beyond their normal proportion in the general Israeli population, in order to increase the statistical power and examine cardiovascular risk factor in this ethnic minority. The multivariable analysis adjusted for ethnicity to overcome this potential confounding.
Moreover, although ISI's that were examined in the study were previously found to show high correlation with the euglycemic insulin clamp (18-21), this method is still considered the gold standard for quantifying insulin resistance. However, this method is invasive and not applicable in large-scale epidemiological studies.
Furthermore, no information on medication or family history were collected during the late 70's intakes.
However, the cohort was mainly composed of healthy and employed subjects. In addition, routine screenings for cancer were not widely used at that time. Finally, the current study did not investigate the association between IR surrogates and cancer site-speci c mortality due to the small number of subjects per group.
The study however present some clear advantages such as the long follow-up over approximately 40 years, the equal representation of both men and women in addition to the representation of an ethnically diverse population. Moreover, blood tests were drawn in the healthy state for research purposes only and analyzed by a single laboratory, avoiding variability in the blood tests analysis. Furthermore, the statistical analysis was performed by two different approaches with similar ndings, reinforcing the study results.

Conclusion
Our ndings demonstrate an association between insulin resistance, as re ected by the lower values of the MCAi, and increased long-term cancer death, in individuals free of diabetes at start of the follow-up.
The MCAi may be considered as a prognostic biomarker in healthy adults.

Declarations
Ethics approval and consent to participate: The Sheba Medical Center Review Board provided approval for this study (approval number 1180). All patients gave their verbal consent to participate in the study during baseline data collection.

Consent for publication:
Not applicable Availability of data and materials: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests:
The authors declare that they have no competing interests Funding: This research received no speci c grant from any funding agency in the public, commercial, or not-forpro t sectors.
Author contributions: YM contributed to the data analysis, the interpretation of data and drafting of the manuscript. RD and AC contributed to the acquisition of the data, to the conception and design of the work, to the data analysis and drafting of the manuscript. DR contributed to the conception and design of the work. DR, AC and RD critically revised the manuscript. All authors have read and approved the nal manuscript.