The Science of Diabetes Self-Management and Care2023, Vol. 49(2) 163 –179© The Author(s) 2023
Article reuse guidelines:
sagepub.com/journals-permissionsDOI: 10.1177/26350106231153073journals.sagepub.com/home/tde
The purpose of this meta-analysis was to examine the association between preexisting diabetes in persons living with cancer on diabetes and oncology-related health outcomes. Understanding this association is of priority because the incidence of both cancer and diabetes mellitus is increasing worldwide.
Methods: A comprehensive review of the literature was conducted in collaboration with an expert health sciences librarian. Two authors independently conducted the screening, data collection, and extraction processes. The risk of bias was assessed using several tools, depending on the study design. Relative risks with 95% confidence intervals were calculated. The alpha threshold was 0.05. All analyses were performed using R statistical software (Metaphor and Demeter packages).
Results: A total of 45 studies met the selection criteria, but 23 were excluded from the synthesis because they did not have the ranked outcome or correct comparison (persons with and without diabetes), totaling 22 studies included in the meta-analysis. In comparison to participants without preexisting diabetes, participants with preexisting diabetes and cancer were found to have a significantly higher risk of infection and cardiovascular, neurological, gastrointestinal, hepatic, and renal complications. Concurrent preexisting diabetes and cancer were also associated with increased health care service utilization and length of hospital stay.
Conclusion: The findings from this review highlight the importance of optimal concurrent management of both diseases by overcoming the compartmentalization of medical specializations through (1) integrated, multidisciplinary, shared, and coordinated clinical care pathways between oncology and diabetes health care providers/teams and (2) the continued development of evidence-based clinical guidelines.
Diabetes mellitus (DM) and cancer are common diseases that significantly impact the health and well-being of persons worldwide. DM is a chronic disease that develops from either the pancreas not producing enough insulin or when the body cannot effectively utilize the insulin it produces.1 Approximately 425 million persons (between 20-79 years of age) are currently living with DM; projections indicate that by 2045, this will rise to 650 million worldwide.2 Cancer is a disease resulting from cellular changes that cause uncontrolled growth and division of cells that affects almost any body part.3 Worldwide, cancer is the leading cause of death, accounting for approximately 10 million persons in 2020.4 For example, in Canada, it is estimated that 1 in every 2 individuals will be diagnosed with cancer and about 1 in 4 Canadians are expected to die from cancer.5,6
Empirical and epidemiological evidence suggests that persons living with DM are at a higher risk of cancer incidence, complications, and mortality in comparison to persons living without DM.7-9 Factors associated with increased cancer-related mortality for persons living with DM include a higher prevalence of infections, increased therapy toxicities, less aggressive cancer-related treatment, and surgical and postsurgical complications.10 Furthermore, due to the heterogeneity of patients’ diabetes-related conditions, diabetes has been found to influence the dose, timing, and therapeutic effectiveness of cancer-related treatment (ie, chemotherapy and neoadjuvant chemoradiation).8,11
The presence of DM-related complications and/or untreated hyperglycemia results in renal, cardiovascular, and neuropathic complications, which are also exacerbated by several anticancer drugs.8 Anticancer drugs have been found to have adverse metabolic effects related to glycemic, lipid, and blood pressure management.8 This is through (a) interfering with the production and/or secretion of insulin and lowering insulin sensitivity, (b) influencing the adipose tissue and liver metabolism through AMP-activated protein kinase activity and modifying lipolysis/lipogenesis, (c) causing vasoconstriction and vasospasm because of the reduction in nitric oxide production and endothelial damage, and (d) the microvascular disease associated with long-term DM limiting the delivery of radio-sensitizing agents to tissues and creating a hypoxic environment that reduces the sensitivity to radiation.8,12 As a result, patients living with both DM and cancer present with unique challenges and complications and may receive either less aggressive cancer-related treatment or diabetes care, both of which may compromise a patient’s health, quality of life, clinical outcomes, and survival.13 The availability of best practice guidelines and protocols regarding the concurrent management of DM and cancer is limited. This is further coupled with the limited knowledge regarding the extent to which the coexistence of preexisting diabetes and cancer impacts short- and/or long-term diabetes and oncology health outcomes and/or complications.8,14 For these reasons, a meta-analysis was conducted to appraise and summarize the evidence regarding the impact and association of preexisting DM, including both type 1 diabetes (T1DM) and type 2 diabetes (T2DM), on diabetes- and/or oncology-related health outcomes for persons living with cancer.
We developed a comprehensive search strategy with an expert health sciences librarian and two of the study investigators (S.I. and D.S.). The databases used to conduct the literature search were MEDLINE, Embase, CINAHL, Psych Info, and Web of Science. The following keywords were used in conjunction with Boolean Operators to refine the literature search: “diabetes” OR “Type 1 Diabetes Mellitus” OR “Type 2 Diabetes Mellitus” OR “glucose intolerance” OR “hyperglycemia” AND “cancer” OR “malignant neoplasm” AND “diabetes outcomes” AND “oncological outcomes.” A hand search of the reference list from the selected articles was also performed.
Studies published between 2000 and 2022 were included if the following selection criteria were met: (a) sample represents persons (ie, children, adults, and/or older adults) living with diabetes (T1DM or T2DM) and cancer; (b) studies used either qualitative designs, randomized control trials (RCTs), controlled clinical trials (CCTs), large cohort studies, and quasi-experimental studies including nonrandomized control studies, pretest and posttest design, interrupted time-series design, and combination design; (c) study reports on diabetes-related outcomes (ie, A1C level) and oncology-related health outcomes; and (d) study reports written in the English language and published between 2000 and 2020. Studies were excluded if (a) the focus was on gestational diabetes, (b) only a description of the relationship of diabetes and cancer was provided, (c) there was no information or report on any diabetes-related and oncological outcomes, and (d) the focus was theoretical and/or conceptual.
To define diabetes- and oncology-related health outcomes, seven health care providers (HCPs) specializing in oncology (ie, oncologists, oncology nurse practitioners and nurses) in Ontario and Quebec, who are part of the circle of care for this unique patient population, were asked to rank the most important health outcomes based on their clinical judgment and expertise. Following completion of the literature review and identifying diabetes and oncology health outcomes, the HCPs were asked to rate on a 9-point Likert scale (1-3 = not important, 4-6 = important but not critical, 7-9 = critical/very important) the outcomes that were perceived to potentially influence diabetes- and/or cancer-related health outcomes. The health outcomes deemed important by the HCPs were related to health care service utilization and diabetes-related clinical complications, specifically: infections (any infectious complication, eg, pulmonary infection, abdominal infection, urinary tract infection, wound infection, sepsis, septicemia, etc), cardiovascular (ie, cardiac arrhythmias, myocardial infarction, hemorrhage, pulmonary embolus), cerebrovascular (ie, stroke), neurologic (ie, neuropathy, neuropathic symptoms, neurotoxicity), gastrointestinal (ie, weight loss, diarrhea, anorexia, late gastro-toxicity, anastomotic leak), wound-related (ie, nonhealing wound, postop fistula, wound dehiscence), hepatic/renal (ie, acute renal failure, hepatic failure, renal insufficiency/failure, jaundice, proteinuria), pain (ie, jaw, muscle, abdominal), and adverse/general health-related (postoperative/surgery complications, postpharmacotherapy complications, hand-foot syndrome, pruritis, aphthous ulcers, etc).
Two researchers (L.M. and S.I.) screened citations for inclusion at both the title, abstract and full text levels. If there was a conflict at the title and abstract level, the citation immediately moved to the full-text level to undergo full review. At the full-text level, conflicts were dealt with through discussion by the reviewers, and where an agreement was not reached, a third reviewer (D.S.) made the final decision.
A standardized data collection form was used to extract the information from the respective studies. The following study characteristics were collected: (a) author’s last name(s) and year published, (b) country in which the study was conducted, (c) research design, (d) sample size, (e) outcomes of interest, (f) estimate of the reported treatment effect (P values, confidence intervals, means, and standard deviations), and (g) overall conclusion of study findings. The following patient- and clinical-related characteristics were collected: (a) sex, (b) mean age, (c) type and the number of comorbidities, (d) type of DM (T1DM or T2DM), (e) duration of diabetes, (f) type of diabetes treatment (eg, oral, insulin, injectables such as GLP-1RA), (g) type of cancer, and (h) duration of cancer diagnosis.
Using valid and standardized forms, we assessed study quality. The Cochrane Risk of Bias Tool was used for RCTs, and the Ottawa-Newcastle (NOS) Risk of Bias Tool was used for cohort and case-control study designs. The quality of each article was independently reviewed by one reviewer (L.M.) and confirmed by a second reviewer (S.I.). Any disagreements between the reviewers were resolved by discussion to reach a consensus. For NOS rating scale (maximum score = 9), the quality of each study was rated using the following score criteria: ≥7 was considered as “good,” 2 to 6 as “fair,” and ≤1 was considered as “poor” quality.
To perform quantitative synthesis of binary outcomes of interest, specifically, DM-related clinical complications (infectious, cardiovascular, neurological, gastrointestinal, wound-related, hepatic-renal, cerebrovascular, and adverse health-related) and increased health care services utilization (prolonged hospital stay, ICU admissions, and 30-day hospital readmissions), we utilized the number of events, proportion, or percentage data. For continuous outcomes of interest (ie, length of hospital stay measured using the number of days), we utilized means with standard deviation data. Meta-analysis of extracted data was performed using DerSimonian and Laird Random Effects approach, and the summary measures of effect were generated as risk ratio (RR) for binary outcomes and mean difference (MD) for continuous outcomes.13 Planned sensitivity and metaregression analyses based on the type of cancer, type of diabetes treatment, length of follow-up, and study quality could not be performed due to too few studies (n < 4) reporting on each type of study-level variable of interest. All analyses were performed using R statistical software (Metafor, and Demetar packages).15-17
The literature search yielded 29 616 titles and abstracts. Of these, 14 511 were excluded because they were duplicates, and 549 were excluded because they were (a) editorials, letters, and/or commentaries; (b) published in a language other than English; and (c) focused on animals. Upon conducting an initial title and abstract review, 10 800 articles were excluded for not meeting the selection criteria. A second title and abstract screening was performed with additional screening questions, resulting in the exclusion of 3710 articles. The Preferred Reporting for Systematic Reviews and Meta-Analysis (PRISMA-2020) flow diagram is presented in Figure 1.18 A total of 47 articles were reviewed in full text, and of those, 45 articles met the inclusion criteria. Of the 45 articles, 22 had sufficient reporting for inclusion in the meta-analysis.19-40 Studies that met inclusion criteria but were not included in the meta-analysis were excluded due to not having ranked outcome or not comparing outcomes between patients with diabetes to those without diabetes (eg, the study included comparisons within the diabetes group only). The DM- and oncology-related health outcomes for persons living with cancer were grouped as (1) health care service utilization; (2) DM-related clinical complications, specifically, infection, cardiovascular, neurological, gastrointestinal, and hepatic/renal; and (3) other.
All included studies were conducted in an acute care setting (n = 22, 100.0%).19-40 The majority of the studies were conducted in the United States (n = 8, 36.4%),19-26 China (n = 3, 13.6%),27-29 Italy (n = 3, 13.6%),30-32 and the Netherlands (n = 2, 9.1%).33,34 Japan, Spain, Turkey, Taiwan, and Germany each contributed one study, respectively.35-39 One study included patient data from multiple international locations, including the United States, Canada, and Brazil.40 The common study design employed was observational (n = 21, 95.5%),19-39 with the majority of these (n = 17)19,21-33,37-39 being retrospective. One included study was an RCT.40 The target population in all studies included were adult participants (n = 22, 100.0%; Table 1).19-40 The majority of the studies were rated as good quality with a median NOS quality rating of 7 (range: 6-8) across the retrospective cohort and case-control studies. One included study based on data from two RCTs were rated as unclear for the overall risk of bias because of lack of information or concerns regarding allocation concealment procedures and industry funding.40
The majority of the studies included participants with both T1DM and T2DM (n = 16, 72.7%)20-26,30-34,36-38,40 or only T2DM (n = 5, 22.7%).19,27-29,35 One study did not report diabetes type.39 The type of cancer varied, with colorectal (n = 6, 27.3%)21,22,31,33,34,40 and breast (n = 3, 13.6%)26,37,39 being the most common, followed by hepatocellular, lung, prostate, gastrointestinal, and head or neck cancers, which all had 2 studies, respectively.19,20,23,25,28-30,32,35,36 Pancreatic24 and esophageal27 cancer each had 1 study. One study included patients with any type of cancer (Table 1).38 The most common intervention/treatment reported in the studies was surgical resection (n = 9, 40.9%)20,23-25,27,29,30,35,36 and chemotherapy (n = 6, 27.3%).22,31,33,37,38 Other treatments used included radiation therapy,19 cytoreductive surgery and chemotherapy,21 kinase-inhibitor pharmacotherapy,32 combined hormonal therapy and chemotherapy,39 and mixed chemotherapy or no chemotherapy.28 One study included patients receiving any form of cancer treatment,26 and 1 study did not report the type of treatment(s) included34 (Table 1). The primary outcomes of interest varied across the studies, with incidence of medical treatment complications being the most reported (n = 9, 40.9%),19,21,22,30,31,33,34,37,38 followed by incidence of surgical and postoperative complications (n = 6, 27.3%),20,21,24,27,35,36 overall survival (n = 6, 27.3%),19,21,27,29,32,40 length of hospital stay (n = 3, 9.1%),25,26,39 cancer recurrence (n = 2, 9.1%),23,29 and mortality (n = 2, 9.1%).21,28 Some studies reported on secondary outcomes, which also varied. A total of 45.5% of the studies did not report on secondary outcomes,19,21-23,31,34,35,37-39 however, of those that did, the most common outcomes were: incidence of surgical and post-operative complications,25,27-29 post-operative mortality,24,36 and length of hospital stay.28,30 Other secondary outcomes included were overall cost of care,20 transfer to another care facility,26 overall survival,27 treatment response,32 cancer progression,32,40 and cancer-specific survival33 (Table 1).
Health care service utilization. The data on health care service utilization were reported by three studies with a total number of 434 persons with both cancer and DM and 5887 persons only with cancer.21,25,33 The evidence from the pooled analysis showed that compared to persons without preexisting DM, persons with cancer and DM were 1.82 times more likely to have high health care service utilization (RR = 1.82; 95% CI, 1.28-2.61; Figure 2). Similarly, for the length of hospital stay, persons with cancer and DM showed a significant increase compared to persons only with cancer (MD = 0.95; 95% CI, 0.29-1.61; Figure 3). These data were reported by seven studies with a total number of 7546 persons with cancer and DM and 47 114 persons only with cancer.20,21,23,26,28,30,39
Infectious complications. The data on infection complications were reported by 10 studies with a total number of 7820 persons with both cancer and DM and 33 005 persons only with cancer.20,21,23,24,27-30,36,39 The evidence from the pooled analysis showed that compared to persons only with cancer, those with cancer and DM were 1.33 times more likely to have infectious complications (RR = 1.33; 95% CI, 1.11-1.60; Figure 4).
Cardiovascular complications. The data on cardiovascular complications were reported by 8 studies with a total number of 7472 persons with both cancer and DM and 32 743 persons only with cancer.20,21,23,24,29,30,32,36 The evidence from the pooled analysis showed that compared to persons only with cancer, those with cancer and DM were 1.66 times more likely to have cardiovascular complications (RR = 1.66; 95% CI, 1.13-2.44) (Figure 5).
Neurological complications. The data on neurological complications were reported by 8 studies with a total number of 4858 persons with both cancer and DM and 29 357 persons only with cancer.20,22,24,31,34,37,38,40 The evidence from the pooled analysis showed that compared to persons only with cancer, those with cancer and DM were 1.48 times more likely to have neurological complications (RR = 1.48; 95% CI, 1.14-1.93; Figure 6).
Gastrointestinal complications. The data on gastrointestinal complications were reported by 6 studies with a total number of 762 persons with both cancer and DM and 4999 persons only with cancer.19,21,27-29,32 The evidence from the pooled analysis showed that compared to persons only with cancer, those with cancer and DM were 2.14 times more likely to have gastrointestinal complications (RR = 2.14; 95% CI, 1.30-3.51; Figure 7).
Hepatic and renal complications. The data on hepatic and renal complications were reported by 3 studies with a total number of 7082 patients with both cancer and DM and 30 852 patients only with cancer.20,21,36 The evidence from the pooled analysis showed that compared to patients only with cancer, those with cancer and DM were 2.21 times more likely to have hepatic and renal complications (RR = 2.21; 95% CI, 1.60-3.06; Figure 8).
Other complications. The evidence from the pooled analysis showed no differences in risk for general health or adverse complications (six studies; RR = 1.26; 95% CI, 0.95-1.67; Figure 9),21,32,33,35,39,40 wound-related complications (four studies; RR = 1.09; 95% CI, 0.73-1.62; Figure 10),20,21,23,24 cerebrovascular complications (two studies; RR = 0.96; 95% CI, 0.56-1.64; Figure 11),20,36 and pain-related complications (two studies; RR = 0.76; 95% CI, 0.51-1.13; Figure 12)32,40 between patients with both cancer and DM and only cancer.
Persons living with cancer and preexisting DM are frequently seen in practice; however, the extent to which DM impacts DM and oncology-related health outcomes are unknown for this patient population. This meta-analysis attempted to address this gap in knowledge by examining the association of preexisting diabetes and cancer impacts on short- and/or long-term diabetes and oncology health outcomes from 22 clinical studies.19-40
Our results demonstrate that compared to persons living only with cancer (and without DM), those with both cancer and DM had a significantly higher risk of experiencing cardiovascular- (e.g., MI, arterial hypertension, and arrhythmia), hepatic- and renal- (e.g., renal insufficiency and failure and jaundice), gastrointestinal- (e.g., nausea, diarrhea, enteric leak/perforation, and late gastrointestinal toxicity), neurological- (eg, neuropathy, neurotoxicity), and infection-related (e.g., wound, sepsis, UTI, pneumonia, and bloodstream) complications. More specifically, it was also found that persons living with cancer and preexisting DM also had longer hospital stays and more health care service utilization (1.82 times more) in comparison to persons without DM. This finding is consistent with that of the literature where it was found that patients with both cancer and preexisting DM had increased cardiovascular- (e.g., myocardial infarction), gastrointestinal-, and infection-related complications and health care service utilization (e.g., emergency department visits).41-45 The increase in acute complications and health care service utilization due to the complex interaction between cancer and DM makes this patient population particularly vulnerable.45 Potential explanation to these findings includes (a) that cancer-related surgeries and treatments interfere with glucose metabolism and increase metabolic stress and/or glycemic decompensation; (b) the biological and pathophysiological mechanisms of DM and cancer (eg, inflammatory pathways, hormone changes, hyperinsulinemia, and hyperglycemia); (c) many anticancer drugs and glucocorticoids have been found to induce hyperglycemia, which increases the risk of for perioperative infections and cardiovascular events; and (d) a possible lack of knowledge of the concurrent management of diabetes and cancer, and/or the deprioritization of diabetes management during the acute phases of cancer treatment.46-51 Interestingly, there were no differences found in cerebrovascular, pain-related, wound-related, or general adverse health events for persons only with cancer. However, it is important to note that for general adverse events, there was a noted inconsistency with the definition of general adverse events across the included studies, potentially influencing an accurate and true representation of the impact and association of this broad-based complication. Additionally, both pain-related and cardiovascular complications were reported only in two studies, which may have limited the representativeness of the potential impact and association of these outcomes. The nonsignificant finding about wound-related complications was of particular interest because chemotherapy and radiation have demonstrated negative effects on wound healing and the processes of such healing are further attenuated with diabetes.52
This unique patient population of those with diabetes and cancer presents with several complex clinical issues that result in diabetes and oncology-related complications and increased health care service utilization. Despite the observed frequency of diabetes and oncology, limited guidelines and protocols exist to guide and inform care delivery processes and treatment regimens for this patient population.53-56 This is further coupled with confusion over the roles and responsibilities of oncology and diabetes/endocrinology HCPs, an overall lack of knowledge surrounding which condition is the pathological source of the symptom(s) and/or presenting complication(s), and the uncoordinated and/or conflicting manner and practices of oncology and diabetes/endocrinology HCPs.46,54 The findings from this meta-analysis highlight a challenging health care period for this unique patient population and have 2 significant implications for the concurrent burden and impact of cancer and DM on patients and the health care system. The first implication is the need for optimal concurrent management of both diseases and overcoming the compartmentalization of medicine and specializations through better interdisciplinary collaboration among oncology and endocrinology HCPs/teams to achieve a shared approach and standard in the level of care afforded to this patient population.53-55,57 A practical solution to these complex issues and addressing and/or mitigating such complications is through the development of integrated, shared, and coordinated clinical care pathways between oncology and diabetes HCPs/teams that place the patient at the center of the care process.54,55,57 Integrated, shared, and coordinated clinical care pathways are novel approaches that hold great potential in overcoming the health disparities in this patient population by (a) increasing communication and coordination between oncology and endocrinology HCPs/teams, (b) taking all diabetes and oncology-related treatment options into consideration and developing tailored treatment plans for patients, (c) sharing evidence-based clinical resources and tools, and (d) educating care teams and patients on the dual disease management by understanding how these diseases interact, the importance of diabetes self-management, and the ability to recognize and address common adverse outcomes/complications.54,55,58-60 Such integrated, shared, and coordinated clinical care pathways have the potential to optimize health care services expenditure and optimize patient outcomes and quality of life, as evidence from preliminary reports,54,55 but, further examination of such interventions is warranted.61
The second implication is the development of evidence-based clinical guidelines and policies for the concurrent management of diabetes and cancer. There are currently a limited number of established standards, guidelines, and/or policies for what should constitute as optimal glycemic levels for these patients coupled with no mandatory baseline assessment of glycemic status for patients at the time of cancer diagnosis.61,62
There are several noted limitations of this meta-analysis. First, in this analysis, we did not assess how the type of diabetes, treatment status of diabetes, and the presence of other comorbidities attenuated these outcomes. There may be key differences in how T1DM and T2DM interact with cancer management and vice versa, which may potentially lead to differences in the risk of complications. Second is the inability to conduct subgroup analysis based on the type of cancer because certain cancers are likely related to higher risk of specific complications. Similarly, we were unable to parse out the differential risk of complications based on the type of intervention (ie, postoperative vs treatment/medication related). Third is due to the heterogeneity across studies based on length of follow-up, diabetes duration, and definition of complications. Finally, the study design of included studies is due to the potential bias associated with the selection of participants and possible confounding due to uncontrolled contextual variables.
Persons living with cancer and preexisting DM are frequently seen in clinical practice. However, the extent the coexistence of preexisting diabetes and cancer impacts short- and/or long-term diabetes and oncology health outcomes. This meta-analysis was conducted to address this knowledge gap. A total of 22 studies were included in this meta-analysis,19-40 with findings that those living with cancer and preexisting DM had a higher risk of infection, neurological, gastrointestinal, hepatic, and renal complications as well as increased health care service utilization and length of hospital stay than those with cancer but without diabetes. The findings from this meta-analysis highlight the need to overcome the compartmentalization of specializations and medicine by (1) improving interdisciplinary collaboration through the development and implementation of integrated, shared, and coordinated clinical care pathways between oncology and diabetes HCPs/teams and (2) continuing development of evidence-based guidelines, protocols, and policies for the concurrent management of cancer and diabetes.
We would like to thank Mrs Donna Fitzpatrick-Lewis for her assistance with the literature search, data screening, and extraction process.
No financial disclosures to disclose.
Lara Murphy https://orcid.org/0000-0002-4543-6591
From Faculty of Health Sciences, McMaster University, Hamilton, Canada (Ms Murphy); Temerty Faculty of Medicine, University of Toronto, Toronto, Canada (Ms Murphy); School of Nursing, McMaster University, Hamilton, Canada (Dr Sherifali); Department of Health Research Methods, Evidence and Impact, Faculty of Health Sciences, McMaster University, Hamilton, Canada (Dr Sherifali, Dr Ali); Diabetes Care and Research Program, Hamilton Health Sciences, Hamilton, Canada (Dr Sherifali, Dr Ali); McMaster Evidence Review and Synthesis Team, McMaster University, Hamilton, Canada (Dr Sherifali, Dr Ali); Lawrence S. Bloomberg Faculty of Nursing, University of Toronto, Toronto, Canada (Dr Ibrahim); The Hospital for Sick Children, Toronto, Canada (Dr Ibrahim); and Centre for Advancing Collaborative Healthcare & Education, University of Toronto, Toronto, Canada (Dr Ibrahim).
Corresponding Author:Sarah Ibrahim, 155 College Street, Toronto, ON M5T 1P8, Canada.Email: sarah.ibrahim@utoronto.ca
