Insulin therapy, for several reasons, is often required in T2D patients with cancer, above all for its efficacy and fast onset of action, being particularly useful in case of severe hyperglycaemia. Secondly, for the flexibility of insulin regimen, helpful to treat the acute and intermittent hyperglycaemia induced by both chemotherapy drugs–especially if cyclically administered–and GCs, often used to reduce the collateral symptoms of chemotherapy, to control the cancer-related pain, or as a component of chemotherapy in blood cancers [9]. Short-acting insulin analogues (lispro, aspart, and glulisine) should be preferred for the management of post-prandial hyperglycaemia in cancer patients. Their fast onset of action and the possibility to inject after meals are useful when, in patients with nausea, vomiting, and difficulty to eat, food intake is not predictable [9, 32].

In patients with also uncontrolled fasting glucose, a strengthening of insulin therapy with a basal-bolus regimen could be necessary. When the introduction of a long-acting insulin analogue is required, insulin degludec and glargine 300 U/mL could be preferred for their greater flexibility in the timing of administration respect to glargine 100 U/mL or detemir [9, 33]. Hypoglycaemia is the main risk of an insulin regimen. An accurate training of the patient and/or his/her caregivers regarding glucose self-monitoring, insulin injection technique, and titration is necessary to reduce this risk. The management of an insulin regimen, especially for patients insulin-naïve or with newly diagnosed hyperglycaemia, could be challenging. Inpatients with a relevant burden deriving from the oncologic disease, the diabetologist should consider the additional load deriving from a complex diabetes therapy.

Studies regarding the carcinogenic effects of long-acting insulin analogues provided uncertain and contrasting results. Due to the lack of conclusive evidence on this topic, the decision to use long-acting insulin analogues in diabetic patients with cancer should be evaluated based on metabolic and non-metabolic risks for each patient [34–37].

Oral insulin secretagogues (sulphonylureas and glinides) stimulate endogenous insulin secretion causing hyperinsulinemia and are, therefore, associated with an increased risk of hypoglycaemia and weight gain.

Sulphonylureas are quite effective in reducing glucose level. The use of short-acting molecules (e.g., gliclazide), useful in the management of post-prandial hyperglycaemia, should be preferred for the lower risk of hypoglycaemia compared to the other molecules of this class (i.e. glibenclamide, glimepiride, and glipizide) [37, 38].

Repaglinide, a meglitinide with short-acting insulin secretagogue effect and different dosage regimen, is useful when meal intake is unpredictable [32].

The evidence concerning the risk of cancer in patients treated with sulphonylureas are conflicting and not univocal. Several observational studies focused on this issue, indicating an increased risk of breast cancer, although mitigated by subsequent, better designed analysis, and an increased cancer-specific mortality, mostly related to the use of glibenclamide [3, 39]. Conversely, sulphonylureas seem to reduce the risk of ovarian and prostate cancer [3].

A molecule acting on post-prandial hyperglycaemia is acarbose, an alpha-glucosidase inhibitor which reduces the absorption of carbohydrates from the small intestine. Acarbose, compared to insulin and secretagogues, has both a lower effect in reducing post-prandial glucose level and a very low risk of hypoglycaemia. Nevertheless, the mild effect on glycaemia and the intestinal AEs (flatulence, abdominal pain, and diarrhoea), limit its use in the management of diabetes in cancer patients [9].

The use of insulin sensitising agents (metformin and pioglitazone) in T2D patients with cancer is supported by the rational to contrast the insulin-resistance often enhanced by antineoplastic drugs and GCs.

A large amount of data is available for metformin, a biguanide widely used in T2D. In addition to the well-known glucose effect, preclinical and clinical studies suggested the anticancer properties of this drug. In vitro, metformin demonstrated an antiproliferative effect by reducing cancer cell proliferation as well as the epithelial-mesenchymal transition, and by inducing apoptosis [39]. A large number of observational studies suggest that metformin reduces the occurrence of several cancers, in particular colon-rectal, breast, prostate, liver, pancreas, gastrointestinal, ovarian cancer [10, 40, 41]. In addition, metformin reduces cancer mortality in comparison with other glucose-lowering agents [42]. Recent evidence from retrospective analysis showed a reduced gastric cancer risk in T2D treated with metformin [43]. Moreover, metformin seemed to increase the radiosensitivity of tumoural tissues: in a meta-analysis of 17 cohort studies, including about 14 thousand patients with different cancer type (prostate, head-neck, rectum, lung, oesophagus, and liver), Rao et al. [44] observed a better response of cancer tissues to radiation therapy and an improved overall survival in patients treated with metformin. Although the evidence, derived from retrospective clinical studies, is not strong enough to recommend metformin in all people with diabetes and cancer, metformin use is reasonable in absence of AEs (gastrointestinal disturbance) or contraindications (chronic kidney failure, contrast medium administration because of the risk of acute nephropathy, and systemic hypoxia) [27, 45].

Another class of oral insulin-sensitising drugs used in T2D are thiazolidinediones (TZDs). Pioglitazone is the only molecule available for human use after the withdrawal of rosiglitazone. These drugs are quite safe in cancer patients, except for a possible increased risk of bladder cancer, observed for pioglitazone in animal studies. Observational studies and meta-analysis did not reach definitive and univocal conclusions. In short, they neither excluded nor confirmed the enhanced risk for bladder cancer [46–49]. Thus, the use of pioglitazone should be carefully avoided in patients with bladder cancer or uninvestigated haematuria [3, 50]. The slow onset of action could be a drawback in patients with acute, severe hyperglycaemia.

The dipeptidyl peptidase-4 inhibitors (DPP4-I) and the glucagon-like peptide-1 receptor agonists (GLP1-RA) are diabetes drugs modulating the incretin system, increasingly used in T2D and potentially useful in cancer patients with diabetes. Preclinical and observational studies showed a possible increased risk of thyroid and pancreatic cancer related to incretin therapies [51–53]. Nevertheless, the evidence of this risk, resulting from studies with methodological critical issues, remained conflicting, inconclusive, and mitigated by subsequent observations [54–56].

The DPP4-I (sitagliptin, saxagliptin, vildagliptin, linagliptin, and alogliptin) have a mild efficacy and a slow onset of action that can reduce their use in patients with moderate-severe and acute hyperglycaemia. The few available findings regarding the influence of DDP4-I on cancer prognosis indicated a safety of these drugs on the risk of neoplasm progression, recurrence, and mortality [57, 58].

The GLP1-RA (exenatide, liraglutide, lixisenatide, dulaglutide, and semaglutide) are more effective in reducing glucose level compared to DPP4-I. Yet, they are related to gastrointestinal side effects (lack of appetite, nausea, vomiting, and weight loss) that are undesirable in patients with cancer.

The newest class of oral diabetes drugs, the sodium glucose transporter-2 inhibitors (SGLT2-I) (empagliflozin, dapagliflozin, canagliflozin, and ertugliflozin), reduces plasma glucose by inhibiting renal glucose reabsorption and increasing urinary glucose excretion. These drugs are not only not linked to an increased risk of malignancies [59, 60], but preclinical studies also indicated, mainly for canagliflozin, an interesting anti-proliferative effect on liver cancer cells [61–65]. SGLT2-I are potentially useful to treat diabetes in cancer patients, although the increased risk of genitourinary infections, volume depletion, and euglycaemic diabetic ketoacidosis should be carefully considered.

The undesirable effect of GCs on glucose homeostasis is well reported and explained by an increased insulin-resistance, liver gluconeogenesis, and a reduced insulin secretion [78, 79] (Table 1). T2D patients using GCs (prednisone, prednisolone, methylprednisolone, and dexamethasone) are likely to develop significant hyperglycaemia, mainly post-prandial, and should be alerted to enhance capillary glucose test checking prior or post lunch/evening meals [27, 71, 80, 81]. In these cases, the use of short-acting insulin analogues represent the first choice and the initial dose of 0.1 U/kg per meal is recommended. Supplementary insulin dosage could be necessary depending on the glycaemic response and the level of pre-prandial hyperglycaemia (0.04 or 0.08 U/kg per meal for pre-prandial glycaemic values between 200–300 mg/dL or above 300 mg/dL, respectively) [82]. In patients already receiving insulin, prandial bolus should be increased according to capillary glucose levels [71]. To overcome the postprandial glycaemic rise of patients on once-daily GCs therapy, the Joint British Diabetes Societies suggest either the morning use of basal insulin (starting dose of 0.4 U/kg) [82, 83], or high dose sulphonylureas in the morning (e.g., gliclazide up to 240 mg in the morning and 80 mg in the evening) [83].

Molecules inhibiting specific components of different signaling pathways may cause hyperglycaemia. The mTOR pathway inhibitors (everolimus, sirolimus, and temsirolimus) used in breast cancer, advanced renal cell carcinoma, and neuroendocrine tumours (NETs) carry out their anticancer effect by inhibiting pathways normally involved in cell metabolism, proliferation, growth, and angiogenesis [84–86]. mTOR-inhibitors increase insulin-resistance and glycaemic levels in both diabetic and non-diabetic cancer patients [71, 87] (Table 1). In particular, everolimus induces hyperglycaemia in 10–15% of cases, therefore requiring glycaemic surveillance [88] (Table 1).

In the last decade, targeted therapies modulating the immune response against cancer cells have acquired a relevant role in the treatment of metastatic melanoma, non-small-cell lung cancer, renal cell carcinoma, bladder carcinoma, head and neck cancer, and lymphomas [72].

Anti-cytotoxic T-lymphocyte 4-antigen antibodies (anti-CTLA-4 Abs), PDL-1 inhibitors, and PD-1 inhibitors could trigger an immune response against pancreatic beta cells and the onset of autoimmune diabetes in about 1% of treated patients [89–91] (Table 1). Hyperglycaemia induced by ICIs is often acute and severe. A meta-analysis of Akturk et al. [89] indicates an onset of hyperglycaemia after a median of 49 days since the first dose of drug, a mean values for glycaemia and HbA1c of 602 mg/dL and 7.9%, respectively, and a diabetic ketoacidosis in 76% of cases. To date, no data are available on glucose trend in T2D patients treated with these drugs. Furthermore, considering the potentially life-threatening onset of ICIs-related hyperglycaemia, a careful clinical and biochemical monitoring is recommended.

Particular attention should be given to patients treated with somatostatin analogues (SST-A) (octreotide, lanreotide, and pasireotide) for acromegaly and NETs. Physiologically, somatostatin (SST) binds pancreatic receptors inhibiting insulin and glucagon release [92]. A dual effect of SST-A on glucose homeostasis has been observed, and this is explained by the different affinity of these drugs for SST receptors (SSTR) [72] (Table 1). Pasireotide strongly inhibits insulin release binding SSTR5 with high affinity. Muhammad et al. [93] reported an increase of glucose values in 88.5% of acromegalic patients treated with pasireotide for 24 weeks, and of HbA1c from 6.1% to 7.8%. Moreover, the percentage of patients with diabetes increased from 32.8% to 68.9% after six months. Otherwise, the treatment with octreotide has been related to hypoglycaemic events because of its preferential activation of SSTR2 and consequent suppression of glucagon production [94].

L-asparaginase, used in patients with acute lymphoblastic leukaemia, reduces insulin secretion depleting L-asparagine in beta cells [9]. Moreover, a pancreatic cytolysis is reported in 7% of cases [95, 96] (Table 1).

Alkylating agents, anti-microtubule agents, antimetabolites, and topoisomerase inhibitors are still widely used, representing the mainstay of many cancer treatments [71]. Their glycaemic effect in T2D patients is not fully understood due to the lack of longitudinal, prospective studies [66, 67]. Understanding of these molecules’ influence on glucose homeostasis is also hindered by the multidrug chemotherapy regimens, often including GCs, which make exploring the role of each drug difficult, highlighting the need of targeted longitudinal trials. The available data indicate the onset of hyperglycaemia in a subset of patients receiving these drugs [66, 71] (Table 1). A systematic review by Hershey et al. [66], including 22 studies of patients with solid cancers, showed that cisplatin (alkylating agents) [97–99], docetaxel and vinorelbine (anti-microtubule agents) [97, 100], 5-fluorouracil (antimetabolites) [98, 101], and doxorubicin (topoisomerase inhibitors) [97] are associated with various degrees of hyperglycaemia in patients previously not affected by diabetes. Nevertheless, these findings were not fully consistent in all studies, as they did not support a causal relationship between the anticancer drug and the occurrence of hyperglycaemia [66]. Our retrospective series of T2D patients with various malignancies showed put forward similar suggestions [67], indicating the same classes of anticancer drugs reported by Hershey et al. [66] among those at major risk to increase HbA1c level.

A negative effect on renal outcomes has been observed in T2D patients treated with various anticancer drugs [67] (Table 2). A significant increase of serum creatinine and reduction of the estimated glomerular filtration rate (eGFR) occurred during the six months following chemotherapy. In about 10% of patients, a detrimental effect was observed also on microalbuminuria. The drug classes more likely to cause renal AEs were mTOR and kinase inhibitors [67]. Everolimus and temsirolimus are known to increase the level of creatinine in up to 10% of treated patients [104]. This is probably due to the hyperglycaemic effect of mTOR inhibitors more than to an intrinsic, detrimental effect of these molecules on renal cells [105, 106] (Table 2).

Tyrosine-kinase inhibitors (TKIs), in particular pazopanib, sunitinib, axitinib, sorafenib, could increase urinary protein levels (asymptomatic albuminuria or nephrotic syndrome) [107, 108]. Similar AEs were reported for bevacizumab and aflibercept, monoclonal antibodies directed against the vascular endothelial growth factor (VEGF), with an incidence between 1% and 38% across clinical trials [108] (Table 2).

Acute interstitial nephritis was described in patients treated with ICIs with occurrence within 3–12 months [108–110] (Table 2).

These findings suggest the need for a clinical (e.g., hydration status, arterial pressure, etc.) and laboratory (creatinine, electrolytes, and urinary protein) monitoring of these patients, in particular in those with nausea and vomiting, to early detect renal AEs [27, 108].

Very limited evidence is available on the influence of anticancer drugs on diabetes retinopathy (Table 2). The 6% of T2D and cancer patients experienced the onset or progression of retinopathy in the six months after chemotherapy, but a correlation with specific anticancer molecules or protocols has not been established [67].

Tamoxifen, an anti-oestrogen widely used in breast cancer, is associated with the occurrence of retinal exudates and haemorrhages [111–113] (Table 2). Some reports described sporadic cases of different types of retinal injury induced by various anticancer drugs (alkylating agents, ICIs, MAP kinase inhibitors, and anti-VEGF) [114–120] (Table 2).

Bortezomib and thalidomide, both used in the treatment of multiple myeloma, can affect the peripheral nervous system by means of poorly understood mechanisms, probably involving the inhibition of angiogenic and neurotrophic factors essential for the nervous system [121] (Table 2). Bortezomib causes neuropathy in up to 40% of patients [122]. Both sensory (paraesthesia, pain) and autonomic (postural dizziness, syncope, diarrhoea, impotence, and urinary disturbances) neuropathic symptoms are described [123]. These AEs are often reversible within six months since bortezomib withdrawal and their incidence decreases with subcutaneous and once-weekly administration [121]. The 70% of patients with prolonged exposure (12 months) to thalidomide experienced bilateral and symmetrical neuropathic sensory, and rarely, autonomic or motor disorders [122] (Table 2).

Both conventional (e.g., anthracyclines, taxanes, and antimetabolites) and novel biological targeted cancer therapies can induce cardiovascular AEs, such as hypertension, QT interval prolongation, left ventricular dysfunction, and heart failure [124–127] (Table 2). These fearful and potentially life-threatening AEs can be irreversible, resulting from classical cytolytic cancer therapies, or reversible, more often related to novel biological therapies [128].

Anthracyclines (doxorubicin, epirubicin, and idarubicin), used both in solid and haematological cancers, are associated with left ventricular dysfunction and heart failure in up to 20% of treated patients. The acute, early onset cardiotoxicity of doxorubicin is dose dependent (doses ≥ 450 mg/m²) and mainly mediated by reactive oxygen species, while late onset AEs are dose-independent and reported in half of the patients within six years [124, 129] (Table 2).

Trastuzumab and TKIs targeting the human epidermal growth factor receptor 2 (HER-2) can lead to hypertension and reduced heart function that, unlike anthracycline-dependent cardiotoxicity, are often reversible [130] (Table 2). A high incidence of hypertension, left ventricular dysfunction, and atherosclerosis was also reported in patients treated with anti-VEGF, multi-target TKIs therapies, and proteasome inhibitors [124, 131] (Table 2).

Although infrequently, cancer immunotherapies could cause heart disease. ICIs are related to cardiomyopathy, myocardial fibrosis, myocarditis, acute heart failure, and lethal myocarditis [124] (Table 2). The risk for ICIs-related cardiovascular AEs increases when these antibodies are used in combination (e.g., ipilimumab and nivolumab) [132].

The early detection of cardiovascular complications of anticancer drugs is mandatory to avoid severe and/or permanent injuries. The use of biomarkers (troponin-I, brain-type natriuretic peptide, and N-terminal-pro-BNP) and imaging techniques (echocardiography, cardiac magnetic resonance) could be useful, although no specific recommendations or guidelines are available [124]. Dexrazoxane, interfering with iron-dependent redox mechanisms, is useful to reduce acute and chronic cardiotoxicity of anthracyclines [133]. However, its use is limited to urgent cases because of the relevant risk of bone marrow suppression [134]. Beta-blockers, angiotensin-converting enzyme inhibitor, angiotensin inhibitors, and mineralocorticoid receptor antagonists gave encouraging results in preventing cardiac injury related to anticancer therapies [135].

Among traditional cardiovascular risk factors, dyslipidaemia should not be forgotten. A close lipid monitoring is necessary, mainly in patients treated with drugs already known to increase lipid level such as those targeting the PI3K-Akt-mTOR pathway, GCs, and anti-androgens [67, 86, 136]. Specific therapies for dyslipidaemia should be evaluated keeping in mind the potential side effects (e.g., cramps in statins users), drug-drug interactions, and patients’ life expectancy [9]. Pravastatin is recommended as first line pharmacological therapy for low density lipoprotein (LDL) reduction, while simvastatin and atorvastatin are contraindicated, because of their interferences with drugs metabolised by cytochrome P450 system [86, 136].