Background

Peripheral artery disease (PAD) and peripheral sensory neuropathy are major lower-limb complications frequently observed in patients with diabetes [1,2,3,4,5]. These are the leading causes of non-traumatic lower-extremity amputation, at least 8 times more common in people with diabetes than in persons without diabetes [6, 7]. Lower-limb complications are associated with major disability, worsening quality of life and huge impacts on health care systems and societies [1, 8, 9]. They are also associated with a dramatic reduction in life expectancy, despite major improvements in medical care during recent decades [3, 10,11,12,13]. The excess risk of premature death observed in patients with diabetes and lower-limb complications may not be completely attributable to cardiovascular disease, as other non-cardiovascular conditions may also be involved [14].

Cancer is a major public health problem worldwide, and is the second leading cause of death [15]. Growing data suggest an excess risk for cancer death among adults with diabetes, compared with others [16,17,18,19]. The risk of cancer increased in people with cardiovascular disease as did the risk of cardiovascular disease in cancer patients [20]. Only few studies have evaluated the risk for cancer death according to vascular complications in people with diabetes, and have reported inconsistent findings [21,22,23]. As far as we are aware, the relationship between lower-limb complications and cancer death has not been investigated in patients with diabetes. In the present study, we aimed to evaluate the incidence and the relative risk of cancer death according to the baseline history of lower-limb complications among patients with type 2 diabetes in the Action in Diabetes and Vascular Disease: PreterAx and DiamicroN Modified-Release Controlled Evaluation (ADVANCE) study.

Methods

Participants

ADVANCE was a large multicentre international randomized trial conducted in patients with type 2 diabetes to test the effects of intensive glucose control using a gliclazide-MR based regimen and blood pressure treatment using a fixed-dose combination of perindopril and indapamide on the incidence of major microvascular and macrovacular events [24]. The design and clinical characteristics of participants in ADVANCE have been published previously [24,25,26]. Briefly, patients with type 2 diabetes mellitus, and at least one additional risk factor for cardiovascular disease, were randomly assigned in a 2 X 2 factorial design to: (i) gliclazide (modified release)–based intensive glucose-control regimen, targeting an HbA1c of ≤ 6.5%, or to standard glucose control, with targets and regimens based on local guidelines, and (ii) a fixed-dose combination of perindopril (4 mg) and indapamide (1.25 mg) or matching placebo. Participants were enrolled between 2001 and 2003; they were seen 3, 4, and 6 months after randomisation, and subsequently, every 6 months until June 2007 (for blood-pressure–lowering comparison) [25]. The follow-up in the randomized glucose-control regimen was continued for an additional 6 months, until January 2008 with a different follow-up schedule as published previously [25, 26]. The protocol of the ADVANCE trial was approved by the Institutional Ethics Committee of each participating centre and all participants provided written informed consent.

Definition of lower-limb complications at baseline

A history of lower-limb complication was defined as the presence at baseline of PAD or peripheral neuropathy. PAD was defined as the presence of at least one of following conditions: lack of peripheral pulse (dorsalis pedis or posterior tibial) palpation, requirement of lower-limb revascularisation (surgery, angioplasty or emergency thrombolysis), or lower-extremity amputation of at least one digit, thought to be due to arterial insufficiency. Peripheral sensory neuropathy was defined as the presence at baseline of at least two neurological abnormalities [27]: disturbance of the light touch sensation, abolition of ankle or knee reflex or chronic (at least 6 weeks) foot ulceration. Data were collected by physicians for each participant based on interview and clinical examination, and reported in the case report forms.

Definition of other conditions at baseline

History of coronary artery disease (CAD) was defined as the presence at baseline of at least myocardial infarction, coronary artery bypass graft, percutaneous transluminal coronary angioplasty, or hospital admission for unstable angina. History of cerebrovascular disease was defined as the presence at baseline of stroke or transient ischaemic attack. Diabetic retinopathy was defined as the presence of at least one of the following conditions: proliferative retinopathy, macular oedema, requirement of retinal laser photocoagulation therapy, or diabetes-related blindness. History of diabetic kidney disease (DKD) was defined as urinary albumin-to-creatinine ratio (ACR) > 30 µg/mg or estimated glomerular filtration rate (eGFR, computed using the Chronic Kidney Disease Epidemiology Collaboration equation) < 60 ml/min/1.73 m2.

Region of origin was categorized as 3 groups: Asia (Philippines, China, Malaysia, and India), established market economies (Australia, Canada, France, Germany, Ireland, Italy, Netherlands, New Zealand, and United Kingdom) and Eastern Europe (the Czech Republic, Estonia, Hungary, Lithuania, Poland, Russia, and Slovakia). Cognitive function was estimated by the mini-mental state examination (MMSE) score and considered as normal (MMSE score ≥ 28) or reduced (MMSE score < 28). Education accomplishment was defined as age at completion of the highest level of formal education, and categorized as basic (≥ 16 years) or low (≤ 15 years).

Primary and secondary endpoints

The primary endpoint for this analysis was cancer death occurring during follow-up. The cause of each death, including cancer-related death, was adjudicated by an independent endpoint committee, blinded to study allocations. The adjudication committee reviewed source documentations for all individuals who died during follow-up. The secondary endpoints were overall and site-specific incident cancers, defined as the first cancer diagnosed after entry to the study according to the International Classification of Diseases Code Tenth Revision (ICD-10). Diagnostic codes used in the study are presented in Additional file 1: Table S1. Incident cancers were collected systematically for all participants during the scheduled study visits from case report forms, and from reports of serious adverse events, without adjudication. Information about the occurrence of all serious adverse events was reported at the time of occurrence between visits. When serious adverse events occurred, the responsible investigator of each centre ensured that the event was reported immediately by completing a Serious Adverse Events Form. The Data and Safety Monitoring Committee regularly reviewed all such events for each centre. The effect of intensive (versus conventional) glucose control on cancer death in ADVANCE has been previously reported [28].

Statistical analyses

Quantitative variables were expressed as mean (SD), or median (25th, 75th percentiles) for those with skewed distributions. Categorical parameters were expressed as numbers and percentages.

Cox proportional hazards regression models were used to estimate hazard ratios (HR) and their 95% confidence intervals (CI) for risk of outcomes according to our three index exposures. We compared the incidence of cancer death according to study allocations: perindopril/indapamide combination (versus placebo) or both intensive glucose control and active blood pressure treatment (versus standard glucose control and placebo group).

Kaplan–Meier curves were used to plot the cumulative incidence of cancer death during follow-up according to the baseline history of lower-limb complication, PAD or peripheral neuropathy at baseline. Survival curves were compared using the log-rank test. Cox proportional hazards regression models were fitted to estimate the risk of outcomes during follow-up according to the history of lower-limb complication at baseline. Analyses were adjusted for baseline age, sex, region of origin and study allocations (model 1), and for model 1 plus any potential confounders: duration of diabetes, body mass index, waist circumference, systolic and diastolic blood pressure, HbA1c, urinary ACR, eGFR, total cholesterol, HDL cholesterol, triglycerides, MMSE score, education accomplishment, history of ever or current smoking, history of past or current alcohol drinking, history of CAD, cerebrovascular disease, diabetic retinopathy or dementia, and use of metformin, insulin, antihypertensive, statins or antiplatelet therapies (model 2). Continuous variables supposed to have U-shaped relationships with death (age, body mass index, waist circumference, systolic and diastolic blood pressure, HbA1c, and eGFR) were introduced within the model with their squared values.

We tested for interaction between PAD and peripheral neuropathy in their association with cancer death. The proportional hazards assumption was checked using the Schoenfeld residuals method.

We performed a series of sensitivity analyses to evaluate the risks of endpoints according to: 1/ the history of lower-limb complications at baseline using an alternative definition of lower-limb complications: the presence of at least one of the following conditions: PAD (lack of peripheral pulse palpation or requirement of lower-limb revascularisation), peripheral neuropathy (at least two neurological abnormalities: disturbance of the light touch sensation, abolition of ankle or knee reflex), or tissue or limb loss (foot ulceration or lower-extremity amputation); 2/ peripheral neuropathy defined alternatively as the presence of at least one neurological abnormality (disturbance of the light touch sensation, abolition of ankle reflex or abolition of knee reflex) or a history of foot ulceration. We evaluated the associations between lower-limb complications and the risk of cancer death in people who have smoked cigarettes regularly (most days for at least a year) and those who have never smoked. Also, we tested the association between lower-limb complications and the risk of cardiovascular and all-cause death. Finally, we used the Fine and Gray method to estimate the subdistribution Hazard ratio for cancer death, while accounting for the competing risk of death from non-cancer causes further to adjusting as in model 2 [29].

Statistics were performed using SAS software, version 9.3 (SAS Institute, Cary, NC, USA, www.sas.com) and Stata software version 15.1 (StataCorp, TX, USA, http://www.stata.com). Two-sided P values < 0.05 were considered statistically significant.

Results

Clinical characteristics at baseline

Among 11,140 ADVANCE participants, 4733 (42%) were women, 4136 (37%) from Asia, 4862 (44%) from established market economies and 2142 (19%) from eastern Europe. Their mean (SD) age and duration of diabetes were 66 (6) and 7.9 (6.3) years, respectively, and the mean (SD) HbA1c was 7.5 (1.6%) at baseline (Table 1). A history of lower-limb complication was reported at baseline in 4293 (38%) participants: 2439 (22%) with PAD and 2973 (27%) with peripheral neuropathy. PAD and peripheral neuropathy were present simultaneously in 1119 (10%) individuals. Characteristics of participants according to a history of PAD or peripheral neuropathy at baseline are displayed in Table 1.

Table 1 Characteristics of participants according to history of peripheral artery disease or peripheral neuropathy at baseline
Full size table

Incidence of cancer death

Cancer death occurred in 316 (2.8%) participants during a median of 5.0 (25th–75th percentile, 4.7–5.1) years of follow-up corresponding to 53,550 person-years and an incidence rate of 5.9 (95% CI 5.3–6.6) per 1000 person-years. The cumulative incidence of cancer death was 2.1% in women and 3.4% in men. The baseline mean (SD) age, duration of diabetes, HbA1c, systolic and diastolic blood pressure of individuals who subsequently died from cancer causes were 69 (6) years, 7.9 (6.6) years, 7.6 (6.6%), 146 (22) mmHg and 80 (11) mmHg, respectively (Table 2). Hazard ratio for cancer death according to each individual parameter is shown in Table 3, after adjusting for study allocations.

Table 2 Characteristics of participants at baseline according to the incidence of cancer death during follow-up
Full size table
Table 3 Risk of cancer death according to each individual parameter at baseline
Full size table

Effects of study treatments on the risk for cancer death

Cancer death occurred in 147 (2.6%) participants assigned to perindopril/indapamide combination versus 169 (3.0%) individuals assigned to placebo (incidence rate: 5.5 [95% CI, 4.6—6.4] versus 6.3 [5.4—7.4] per 1000 person-years) with no significant difference between blood pressure treatment groups: HR 0.86 (95% CI, 0.69–1.08). Cancer death occurred in 65 (2.3%) participants [incidence rate 4.8 (3.8–6.2) per 1000 person-years] randomized to both intensive glucose control and active blood pressure treatment versus 86 (3.1%) individuals [incidence rate 6.4 (5.2–8.0) per 1000 person-years] assigned to standard glucose control and placebo group [HR 0.75 (0.54–1.03)].

Risk of cancer death according to history of lower-limb complications at baseline

The cumulative incidences (Table 4 and Fig. 1) and incidence rates of cancer death were higher in participants with a baseline history of lower-limb complications [8.3 (95% CI, 7.2–9.7] vs. 4.4 (3.7–5.2) per 1000 person-years], PAD [8.9 (7.3–10.8) vs. 5.1 (4.4–5.8) per 1000 person-years] or peripheral neuropathy [8.4 (7.1–10.1) vs. 5.0 [4.3–5.7] per 1000 person-years), compared with individuals without this complication. The risk of cancer death was higher in patients with a baseline history of lower-limb complication [HR 1.62 (1.29–2.04), p < 0.0001], PAD [1.43 (1.12–1.83), p = 0.004] or peripheral neuropathy [1.49 (1.19–1.88), p = 0.0006], compared with those without these conditions after adjusting for age (and its square), sex, region of origin and study allocations. These associations remained significant after further adjustment for a wide range of potential confounders (model 2, Table 4). When we used our alternative definition of lower-limb complication, PAD and peripheral neuropathy remained associated with increased risk of cancer death, but not the history of tissue or limb loss (Additional file 1: Table S2). Comparable results were also observed when peripheral neuropathy was defined as the presence of at least one neurological abnormality or a history of foot ulceration (Additional file 1: Table S3). Associations between lower-limb complications and risk of cancer death were comparable when were stratified analyses by smoking status, with no evidence for statistical interaction (Additional file 1: Table S4). No statistical interaction was observed between PAD and peripheral neuropathy in their association with cancer death (p = 0.14).

Table 4 Risk of cancer death according to history of lower-limb complications at baseline
Full size table
Fig. 1
figure 1

Cumulative incidence of cancer death during follow-up according to history of lower-limb complications at baseline. Solid line: absence of complication; dashed line: presence of complication. P values (log-rank test) for comparing absence and presence < 0.0001 for all three complications. LLC lower-limb complications, PAD peripheral artery disease

Full size image

Risk of cancer death according to baseline history of lower-limb complications considering death from non-cancer causes as a competing risk

All-cause and cardiovascular death occurred in 1031 (9.2%) and 542 (4.9%) participants during follow-up. Their incidence rates were 19.2 (95% CI 18.1–20.5) and 10.1 (9.3–11.0) per 1000 person-years, respectively. We observed an increased risk of all-cause and cardiovascular death in patients with a baseline history of lower-limb complications or PAD, compared with those without these complications (Additional file 1: Table S5). The history of peripheral neuropathy was associated with increased risk of all-cause, but not cardiovascular, death (Additional file 1: Table S5). The association between lower-limb complications and cancer death did not change substantially when we considered death from non-cancer causes as a competing risk (Additional file 1: Table S5).

Risk of incident cancers according to history of lower-limb complications at baseline

Incident cancers were diagnosed in 700 (6.3%) participants during follow-up (Additional file 1: Table S1), corresponding to 52,473 person-years and an incidence rate of 13.3 (95% CI, 12.4–14.4) per 1000 person-years. The cumulative incidences (Table 5) and incidence rates of cancers were higher in participants with a baseline history of lower-limb complication [17.1 (95% CI, 15.4–19.0) vs.11.0 (9.9–12.2) per 1000 person-years], PAD [18.9 (16.5–21.6) vs. 11.8 (10.8–12.9) per 1000 person-years] or peripheral neuropathy [16.7 (14.7–19.0) vs. 12.1 (11.1–13.3) per 1000 person-years], compared with those without these conditions. The relative risk of incident cancers was higher in participants with (vs. without) a baseline history of lower-limb complications [HR 1.23 (1.05–1.44), p = 0.01] or PAD (1.20 (1.01–1.43), p = 0.04], but not those with peripheral neuropathy [1.14 (0.97–1.35), p = 0.11] after adjustment for cofounding variables as in model 2 (Table 5). Association with PAD was particularly observed for solid and digestive malignant neoplasms (Additional file 1: Table S6). Based on the alternative definitions of lower-limb complications, PAD remained associated with excess risk of incident cancers, while peripheral neuropathy and tissue or limb loss did not (Additional file 1: Table S2). The history of diabetic neuropathy, defined as the presence of at least one neurological abnormality or a history of foot ulceration, was not associated with the risk of incident cancers (Additional file 1: Table S3).

Table 5 Risk of incident cancers according to history of lower-limb complications at baseline
Full size table

Discussion

In the present study, we investigated the relationship between a baseline history of lower-limb complications and the risk of cancer death in patients with type 2 diabetes in the ADVANCE study. We observed an independent association between the history of lower-limb complications, both PAD and peripheral neuropathy, and excess 5-year risk of cancer death. No statistical interaction was observed between PAD and sensory peripheral neuropathy in their association with the risk of cancer death. Our findings were reliable when we considered death from non-cancer causes as a competing risk. Of note, lower-limb complications and PAD were associated with increased risk of both all-cause and cardiovascular death, while neuropathy was associated with only all-cause death. The history of PAD, but not peripheral neuropathy, was also associated with a higher incidence of cancers, mainly solid and digestive neoplasms.

Lower-limb complications and risk of cancer death

As far as we know, this is the first report of increased risk of cancer death related to lower-limb complications in patients with type 2 diabetes. Few previous studies reported increased incidence of cancers among patients with intermittent claudication or critical limb ischemia in general population [30, 31]. A recent study has shown an association between lower-limb arterial thrombosis and increased risk of subsequent cancers in 6600 patients (12% with diabetes) from Danish nationwide population-based registries [32]. A lower-limb arterial thrombosis was also associated with a high risk of all-cause mortality following some site-specific (mainly smoking-dependent) cancers.

Potential mechanisms linking lower-limb complications to cancer death

Our findings cannot allow any etiological conclusion, and may only suggest lower-limb complications as proxy for cancer death. Indeed, lower-limb complications are surrogates for advanced microvascular and macrovascular disease, and may lead to a greater likelihood of multisystem disease. Nevertheless, lower-limb complications and malignancies may share some modifiable risk factors, including age, obesity and tobacco smoking [33,34,35]. These risk factors are unlikely to explain our findings as the key observed associations were reliable after adjustment for these confounders. Furthermore, the magnitude of the associations between lower-limb complications and cancer death were comparable when we performed analyses in people who have smoked cigarettes regularly and those who have never smoked, without evidence for statistical interaction. Further studies are needed to investigate the potential contribution of smoking in the relationship between lower-limb complications and cancer death. Lower-limb complications and malignancies may also share some common biological disorders including chronic inflammation, increased advanced glycation endproducts and oxidative stress [36,37,38,39]. Oxidative stress leads to lipid peroxidation and DNA damage, which are involved in microvascular disease, atherosclerosis and cancer [36, 40]. Furthermore, endothelial dysfunction, an important impairment involved in lower-limb complications, has also been linked to increased risk of solid-tumor cancer [41]. While lower-limb complications, involving microvascular disease and atherosclerosis (as a result at least partly of early endothelial dysfunction), cancer is mainly characterized by excessive angiogenesis, which may be driven by endothelial metabolic aberrations [42].

Effects of glucose and blood pressure controls in risk of cancer death

We did not observe an association between ADVANCE allocation to blood pressure treatment and the risk of cancer death, as was previously reported for intensive glucose control [28]. However, we have observed a non-significant decrease in risk of cancer death in participants assigned to both glucose control and blood pressure interventions, compared with those allocated to conventional glucose control and placebo blood pressure treatments. This observation may raise the hypothesis wondering the potential effect of multifactorial intervention on the incidence of cancer death in people with type 2 diabetes. Randomised clinical trials are required to address this hypothesis. Of note, Rasmussen-Torvik and coworkers reported an inverse association between the number of ideal cardiovascular health metrics, as defined by the American Heart Association [43] (including blood pressure and glycemia) and combined cancer incidence in the Atherosclerosis Risk In Communities (ARIC) cohort [44].

Strengths and limitations

The main strength of our work is the use of a large international multicentre study of 11,140 patients with type 2 diabetes, with a comprehensive range of data including demographic, clinical and biological features at baseline. Participants were also prospectively followed for 5 years, with adjudicated causes of mortality, including cancer death within a central and independent adjudication committee process. However, incident cancers were collected from case report forms and reports of serious adverse events, without adjudication or validation using cancer registry data. The other limitation of our study is related to the post hoc analyses of a randomized controlled trial, and the use of a clinical trial population, which may not be representative of all patients with type 2 diabetes. Also, we used a pragmatic assessment of lower-limb complications without a prespecified research protocol, and we missed some data regarding PAD (intermittent claudication, ankle brachial index, toe pressure or transcutaneous oxygen pressure) or peripheral sensory neuropathy (symptoms or vibration sensation test). We do not believe that these issues influenced our findings as we investigated a widespread spectrum of data including clinical examination (pulses palpation, light touch sensation and reflex tests), condition (foot ulceration) and procedures (revascularisation and lower-extremity amputation). However, we cannot exclude that non-diabetic causes of neuropathy could have biased our findings, especially neurotoxicity of cancer treatment (chemotherapeutic drugs and radiotherapy), paraneoplastic neurologic disorders or nervous dysfunction caused by the cancer [45]. Nevertheless, cancer-related neuropathy seemed to be unlikely, since we dealt with a high prevalent neuropathy (n = 2973, 27%) consistent with diabetic origin, while incident cancers were diagnosed only in 700 (6%) participants. Finally, our findings were not controlled for socioeconomic position as we missed appropriate data, except for education accomplishment.

Conclusions

In summary, our study links PAD and sensory peripheral neuropathy to excess 5-year risk for cancer death in patients with type 2 diabetes. In addition, PAD was also associated with increased risk of incident cancers, especially solid and digestive ones. Our findings provide new evidence on the non-cardiovascular prognostic burden of lower-limb complications in people with type 2 diabetes, and encourage continuing careful evaluation and management of cancers in this population.