Evaluation of alpelisib-induced hyperglycemia prophylaxis and associated risk factors in PIK3CA-mutated hormone-receptor positive, human epidermal growth factor-2 negative advanced breast cancer

SOLAR-1 investigated alpelisib-fulvestrant (ALP + FLV) in patients with HR + /HER2-, PIK3CA-mutated advanced breast cancer and demonstrated a clinically significant increase in all-grade and grade (G) 3-4 hyperglycemia (HG) compared to placebo-fulvestrant. Given high rates of HG, a preventative protocol and identification of associated risk factors was implemented. This single-center, retrospective study included patients receiving ALP + FLV. One week before ALP initiation, patients started an insulin-sensitizer. Patients had fasting plasma glucose (FPG) levels drawn day 8, 15, 28, then monthly. Primary outcome was incidence of G2-4 HG by day 28. Risk factors assessed included age, BMI, FPG, and HbA1c. Number of risk factors were compared between patients with and without HG. Sixteen women were included with median age of 59 years. The cohort was 69% White, 25% Black, 75% with BMI ≥ 25 kg/m2, and 50% with history of diabetes. By day 28, 9 patients (56%) had G2-4 HG, with only 3 (19%) G3 and zero G4. Patients with G2-4 HG had a median of 2 risk factors compared to only 1 if no HG (p = 0.03). 5 patients (31%) required a temporary hold of ALP and 3 (19%) required dose reduction due to HG. 13 patients permanently discontinued ALP—9 due to disease progression and 4 from an adverse event (only 1 HG). Implementation of a HG prophylaxis protocol with ALP in a single-center study demonstrated fewer G3-4 HG events compared to that seen in SOLAR-1 (19% vs 36.6%). An increase in HG-associated risk factors correlated with a higher incidence of G2-4 HG.


Introduction
Breast cancer remains the most commonly diagnosed cancer among women in the USA, with hormone receptor-positive (HR +) and human epidermal growth factor 2-negative (HER2-) being the most common subtype accounting for over 70% of all cases [1,2]. First-line therapy for patients with metastatic HR + /HER2-breast cancer generally consists of a combination of endocrine therapy with a CDK4/6 inhibitor [3]. However, approximately 40% of patients also demonstrate a mutation in PIK3CA, the gene encoding the p110α subunit of phosphatidylinositol-3-kinase (PI3K) [4]. PI3Ks are a family of kinases that play a critical role in regulating biological functions, including cell growth, protein synthesis, angiogenesis, immune function, and cellular metabolism [5]. Aberrant activation of the PI3K pathway has been linked with endocrine therapy resistance, making a targeted PI3K inhibitor, alpelisib (ALP), desirable [5].
ALP, a small-molecule PI3K inhibitor, is selective for the p110α isoform and is approved in combination with fulvestrant (FLV) in patients with advanced HR + /HER2-, PIK3CA-mutated breast cancer following progression on or after endocrine-based therapy [6]. This approval is based on the findings of the phase III SOLAR-1 trial, which demonstrated an improved overall response rate and a significantly longer progression-free survival compared to placebo (P) + FLV [7].
While the efficacy data is promising, ALP does have associated toxicities-most notably hyperglycemia (HG) and rash. In SOLAR-1, patients receiving ALP + FLV saw a clinically significant increase in grade 3 or 4 HG (36.6% vs. 0.7%) and grade 3 rash (9.9% vs. 0.3%), when compared to P + FLV [7]. Alpelisib-induced HG is likely related to the crucial role that p110α plays in mediating cellular response to insulin [5]. Inhibition of p110α blocks insulin signaling, leading to glycogenolysis in the liver and decreased glucose uptake into skeletal muscle and adipose tissue, ultimately resulting in a state of insulin resistance, hyperglycemia, and hyperinsulinemia [8]. As investigators observed an increase in HG and rash in the ALP treatment arm, the study protocol was amended to restrict inclusion of patients with diabetes and provide adverse event management recommendations [7,9].
While HG may be prolonged in patients with any degree of insulin resistance, current guidance recommends starting an insulin-sensitizing anti-hyperglycemic agent, such as metformin, a thiazolidinedione, or a sodium glucose cotransporter-2 (SGLT-2) inhibitor, at onset of hyperglycemia [9,10]. Additionally, a mouse model demonstrated an improved efficacy/toxicity ratio of PI3K inhibitors with the use of metformin, an SGLT-2 inhibitor, and dietary adjustments [10]. ALP-induced HG is an expected adverse event, and patients being treated in the non-clinical trial setting may possess factors that increase HG risk. Thus, a preventative strategy was implemented within the Duke Breast Oncology Clinic (DBOC) to minimize HG, with the goal of limiting dose reductions and drug discontinuation. Here we present the results of our primary analysis in the interest of establishing an optimal strategy for minimizing ALPinduced HG and associated toxicities.

Design and setting
This IRB approved, single-center, retrospective, cohort study was conducted to assess the HG prevention strategy implemented in patients receiving ALP within DBOC. We identified 24 patients aged 18 years or older with HR + / HER2-, locally advanced or metastatic breast cancer with a documented PIK3CA mutation who were prescribed ALP between June 2019 and April 2021, in combination with FLV. Patients were excluded if they did not complete at least one cycle (28 days) of ALP or if they were seen as a consult or second opinion for another institution.
This pharmacist-driven HG prevention protocol created and utilized in DBOC starting June 2019 includes a baseline assessment of risk, implementation of anti-hyperglycemic prophylaxis, patient education from a pharmacist and dietician, and close follow-up with pharmacist-monitoring for occurrence of HG or other related adverse events. Baseline labs were obtained, including a fasting plasma glucose, HbA1c, complete metabolic panel, and complete blood count with differential. One week prior to starting ALP, patients initiated metformin 500 mg daily (extended release formulation). Additional or alternative insulin-sensitizing agents, including pioglitazone (a thiazolidinedione) or canagliflozin (an SGLT2 inhibitor), were considered for metformin intolerance. Patients were allowed to have prior anti-hyperglycemic exposure or a history of well-controlled type 2 diabetes mellitus (T2DM), gestational diabetes, or pre-diabetes. Patients were required to demonstrate a HbA1c < 6.5% or consistent fasting plasma glucose (FPG) readings ≤ 140 mg/dL for at least one week prior to ALP initiation, which is representative of controlled diabetes according to the American Diabetes Association (ADA) guidelines [11]. If necessary, further blood glucose control could be achieved through modification of the patient's current antihyperglycemic regimen through addition of a new agent or dose optimization of an existing agent. Patients were classified as diabetic if previously diagnosed by a healthcare provider and pre-diabetic if they had a HbA1c ≥ 5.7% prior to initiating ALP.
All patients were prescribed and instructed on the use of diabetic testing supplies, including a glucometer, test strips, and lancets. At home, patients were instructed to check fasting blood glucose (FBG) levels twice weekly or more often as clinically indicated. Patients were seen in clinic by a provider or pharmacist on days 8, 15, and 28 or were contacted by phone if the patient elected to have external lab checks on days 8 and 15 to assess for presence of HG or other adverse events. If FPG levels remained stable after day 28, patients continued with monthly labs and clinic visits and a repeat HbA1c check every 3 months.
Additionally, all patients were prescribed antihistamine prophylaxis with cetirizine 10 mg daily to prevent rash, starting one week prior to ALP initiation given previous additional data analyses from SOLAR-1 that suggested reduced cutaneous toxicity (rash) for patients receiving antihistamines. During each clinic visit or phone call assessment, patients were assessed for newly developed adverse events, including fatigue, nausea, vomiting, diarrhea, swelling/ edema, increased transaminases over 3 times the upper limit of normal, increased serum creatinine over 1.5 times the patient's baseline, pancreatitis, pneumonitis, and mucositis.

Data source and data collection
Data were collected for all patients including demographics, oncologic history, and clinical data pertaining to ALP monitoring and safety through retrospective chart review. Data were collected and managed using Research Electronic Data Capture (REDCap) electronic data capture tools hosted at Duke University [12,13]. PIK3CA status was assessed using next-generation sequencing. Routine clinical and laboratory assessments, including glucose and symptom management, were conducted at baseline, day 8, day 15, day 28, then every four weeks or as clinically indicated. A 3-to-7-day window was allowed during the first month for lab collection and a 10-day window was allowed for subsequent monthly lab collections for patient convenience. Patients were contacted intermittently at the provider's or pharmacist's discretion for more frequent monitoring. Adverse events were assessed continuously over a 6-month period following ALP initiation or throughout the duration of the study period (April 2, 2021), whichever came first.

Measures
The primary outcome was the incidence of grade 2-4 HG, defined as a FPG ≥ 160 mg/dL, within 28 days of ALP initiation. Secondary outcomes included the incidence of grade 3-4 HG within 28 days of ALP initiation; grade 2-4 HG at day 8 and day 15; all-grade rash within 28 days of ALP initiation; as well as all-grade rash within 6 months. In this study, adverse event grading was classified according to Common Terminology Criteria for Adverse Events (CTCAE) version 4.03 [14]. The overall exposure to ALP as well as the incidence and reasons for dose reductions, therapy interruptions, and discontinuations were assessed. Patient-specific risk factors were evaluated to determine if certain risk factors or if the number of risk factors present were associated with increased incidence of HG development. Risk factors that were evaluated include age ≥ 65 years, BMI at baseline ≥ 25 kg/m 2 , baseline FPG ≥ 100 mg/dL, and baseline HbA1c ≥ 5.7%. Patients were also reviewed for presence of comorbidities, including hypertension, hyperlipidemia, and cardiovascular disease such as heart failure with ejection fraction < 50% or arrhythmia.

Statistical analysis
Descriptive statistics were used to summarize patient demographics and clinical characteristics. Primary and secondary outcomes related to HG were reported as the proportion of patients experiencing the event with the exact binomial 95% confidence interval using the Wilson method. To evaluate possible predisposing factors for grade 2-4 HG, risk factors were summarized and the number of risk factors were compared using Wilcoxon rank sum tests for those who experienced grade 2-4 HG and those who did not. All statistical analyses were conducted using R 4.0.0 (R Core Team, Vienna, Austria). The level of significance was set at 0.05.

Patient characteristics
Between June 19, 2019 and April 2, 2021, a total of 16 women with HR + /HER2-advanced breast cancer with a PIK3CA mutation were newly initiated on ALP therapy at DBOC and were subsequently included in the analysis. The demographic characteristics of the patients in the cohort are detailed in Table 1. Median age was 59 years (range 39-75), and the cohort was 69% White, 25% Black, and one patient was Native American. Patients were predominantly overweight or obese (44% BMI 25-29.9 kg/m 2 , 31% BMI ≥ 30 kg/m 2 ). All patients were post-menopausal. Three patients (19%) had a history of diabetes, two with type 2 diabetes mellitus (T2DM) and one with gestational diabetes during a prior pregnancy. Five patients (31%) had pre-diabetes based on HbA1c 5.7-6.4% at baseline. Median FPG of the cohort at baseline was 102 mg/dL (range 85-143), and median HbA1c at baseline was 5.6% 1 3 (range 4.2-6.5). Patients were also evaluated for presence of comorbidities, including hypertension (7 patients, 44%), hyperlipidemia (1 patient, 6%), and cardiovascular disease such as heart failure with ejection fraction < 50% or arrhythmia (2 patients, 13%). Six patients (38%) had one comorbidity; two patients (13%) had two comorbidities; and eight patients (50%) had no comorbidities present. Additionally, all patients had an ECOG performance status ≤ 2, with the majority having an ECOG performance status of 0-1.

Alpelisib treatment
At the data cutoff (April 2, 2021), three patients (19%) continued ALP therapy. The median duration of therapy among those who had discontinued was 86 days (range 24-442). The most common reason for discontinuation of ALP was progression of disease in 9 patients (56%), with only one patient discontinuing therapy due to hyperglycemia and three (19%) discontinuing due to another adverse event. Dose interruptions for ALP occurred in 10 patients (63%), with the most common reasons being hyperglycemia in 5 patients (31%) and rash in 3 patients (19%). The longest duration of dose interruption was 64 days in a patient receiving palliative radiation therapy. Dose reductions occurred in 7 patients (44%), with the most common reason being hyperglycemia in 3 patients, followed by nausea (n = 2), rash (n = 1), and hepatic dysfunction (n = 1). Alpelisib exposure is further explained in Table 3.

Hyperglycemia
Nine patients (56%, 95% CI: 33%-77%) experienced the primary outcome of grade 2-4 HG by day 28, with a median time to onset of 7 days (range 3-17 days). Six of these patients (38%) experienced grade 2 HG, three patients (19%) experienced grade 3 HG, and zero patients experienced grade 4 HG by day 28. No additional patients developed grade 2-4 HG after day 28. Fifteen patients (94%) received metformin as their primary anti-hyperglycemic prophylaxis, though two patients did not start metformin until they initiated ALP. One patient received pioglitazone 15 mg/day as primary prophylaxis due to an inability to tolerate metformin initially. For the two diabetic patients receiving anti-hyperglycemic therapy prior to ALP consideration, one patient was receiving metformin alone and one patient was receiving both metformin and a DPP4 inhibitor for diabetes management; both patients continued these therapies at time of ALP initiation. Ten patients (63%) required a dose escalation of their antihyperglycemic agent throughout the duration of the study period. Six patients (38%) required at least one additional or alternative agent (one metformin and canagliflozin, three pioglitazone, one glipizide XL, and one insulin aspart and insulin glargine), with only four patients (25%) requiring two or more therapies at a single time.

Risk factor assessment
Patient-specific risk factors were stratified by whether or not patients experienced grade 2-4 HG by day 28 after ALP initiation. Those with grade 2-4 HG were older (median age 61 years vs. 50 years), had a higher BMI (median 30 kg/m 2 vs. 22 kg/m 2 ), had a higher baseline FPG (median 107 mg/ dL vs 93 mg/dL), and had a higher baseline HbA1c (median 5.7% vs 5.4%), compared to those who did not experience grade 2-4 HG. Additionally, when patients were assessed based on the number of risk factors that they met, patients with grade 2-4 HG had a significantly higher number of risk factors compared to those who did not (median 2 vs. 1, p = 0.03).

Adverse events
Five patients (31%) experienced rash by day 28 despite antihistamine prophylaxis with cetirizine, and one additional patient experienced rash by 6 months. The median time to rash onset was 13 days (range 11-74 days). Four patients (25%) experienced grade 2 rash, while two patients (13%) experienced grade 3 rash.
Additional adverse events were evaluated at day 8, 15, and 28-including fatigue, nausea, vomiting, diarrhea, swelling/edema, increased transaminases over 3 times the upper limit of normal, increased serum creatinine over 1.5 times the patient's baseline, pancreatitis, pneumonitis, and mucositis. Nine patients had a documented adverse event by day 8, with an additional 5 patients by day 15. The most common adverse event experienced by day 28 was fatigue (69%), followed by diarrhea (63%), nausea (56%), and mucositis (31%).

Discussion
To our knowledge, this study represents the largest analysis of ALP with a focus on minimizing hyperglycemic events and is the first to examine clinical treatment patterns and safety outcomes outside of the SOLAR-1 trial. The key findings of our study are: (1) the lower rate of grade 3-4 hyperglycemic events seen from ALP with the use of a preemptive anti-hyperglycemic protocol compared to rates cited in previous studies [7,15] and (2) the significantly higher number of risk factors in patients prescribed ALP who ultimately experienced grade 2-4 hyperglycemia compared to those who did not.
Identification of risk factors for hyperglycemia is critical due to the risk of toxicity from high glucose levels, which can present most critically as acute hyperglycemia with resulting ketoacidosis and may warrant an emergency room visit or even admission into the hospital for glucose control. Additional concerns include both microvascular and macrovascular diseases if not managed quickly -ranging from cardiovascular and cerebrovascular pathologies to neuropathy, nephropathy, and retinopathy [16,17]. A detailed safety analysis of the SOLAR-1 trial elaborated on the baseline characteristics and outcomes of patients experiencing hyperglycemia [9]. Increases in FPG were more pronounced and the incidence of all-grade and grade 3-4 HG was higher in the pre-diabetic and diabetic subgroup compared to those who were euglycemic at baseline [9]. By evaluating ALP use in a non-clinical trial patient population, we included a large proportion of patients with pre-disposing risk factors for hyperglycemia (i.e. controlled T2DM, pre-diabetes, obesity, elevated baseline FPG or HbA1c, hypertension, hyperlipidemia, and cardiovascular disease), which are all prevalent in the Southeast region of the USA. In our patients, implementation of a pre-emptive hyperglycemia prevention protocol in patients being prescribed ALP demonstrated fewer grade 3-4 hyperglycemic events (19%) compared to rates reported in SOLAR-1 (36.6%) and BYLieve (28%), indicating that even high-risk patients may be eligible for ALP therapy if prevention is optimized [7,15]. However, a larger prospective study implementing prophylactic antihyperglycemics is warranted to confirm this finding.
Based on this retrospective study, it is our recommendation that patients being considered for ALP be started on anti-hyperglycemic prophylaxis with an insulin-sensitizing agent one week prior to initiating ALP, with metformin extended-release (XR) being an ideal first-line option due to ease of access, low cost, and increased tolerability of the XR formulation and low risk of hypoglycemia [18]. In patients unable to tolerate metformin, a thiazolidinedione or SGLT2 inhibitor are both optimal second-line options as their mechanism still serves to increase insulin sensitivity, with choice being dependent of patient-specific factors and preference [18]. Patients already receiving anti-hyperglycemic therapy for pre-diabetes or type 2 diabetes mellitus should continue on anti-hyperglycemics, and consideration should be given to dose titration and/or addition of an adjunctive agent prior to starting ALP consistent with management of ALP-induced hyperglycemia and its understood mechanism.
Metformin is generally considered to be a good firstline option. However, it demonstrated only a minor benefit on HG reduction in mouse models receiving ALP, while SGLT-2 inhibition was associated with a more profound decrease in HG and hyperinsulinemia [10]. For higher risk patients, SGLT-2 inhibition could be considered in the first line setting or in combination with metformin in higher risk patients. However, more research is needed to provide definitive recommendations in these populations and to further elucidate what patient-specific factors may increase risk of alpelisib-induced hyperglycemia.
Optimized blood glucose control may also serve to improve efficacy outcomes with ALP if dose intensity is maintained. The BYLieve study demonstrated relatively longer progression-free survival in patients who received a higher median dose intensity of ALP, although patients receiving high-dose (≥ 248 mg/day) and low-dose (< 248 mg/day) ALP both demonstrated improved PFS over placebo [9]. By optimizing HG prevention strategies initially and minimizing the occurrence of high-grade HG, we anticipate fewer dose reductions, therapy interruptions, and discontinuations. This may correlate with improved efficacy outcomes though future research in this area is necessary.
Endocrine therapy is the therapeutic backbone for managing HR + /HER2-advanced or metastatic breast cancer and is recommended first-line in combination with a CDK4/6 inhibitor [3]. Notably, only a small subset of the SOLAR-1 population received previous treatment with a CDK4/6 inhibitor, and when stratified based on prior lines of therapy, a lack of prior CDK4/6 inhibitor treatment was associated with a significant improvement in progressionfree survival in patients with a PIK3CA-mutation receiving ALP + FLV [7]. In our population, all but one patient prescribed ALP received a prior CDK4/6 inhibitor and all patients continued FLV from prior therapy or initiated FLV prior to starting ALP, which is representative of real-world clinical practice. The phase II trial, BYLieve, is investigating the therapeutic efficacy of ALP + FLV in patients after disease progression on a CKD4/6 inhibitor to further answer this question of sequencing. While our retrospective study did not specifically evaluate these outcomes, our data does reflect feasibility and benefit of ALP + FLV post-CDK4/6 inhibitor therapy.
Although our sample size is relatively small, it is representative of one of the only patient populations receiving ALP in combination with prophylactic antihyperglycemic therapy. Our study is limited by its observational and retrospective nature and lack of a control group which does not allow for direct comparison of hyperglycemic events observed in SOLAR-1. We were not able to control for concomitant dietary changes, although all patients did receive a referral to nutritional counseling for dietary modifications, or for patient adherence to the preventative anti-hyperglycemic regimen or prophylactic antihistamine for rash prevention. Additionally, implementation of patient education including dietary recommendations as well as more intensive monitoring could also have been contributory to results seen. Some patients also elected to have labs obtained externally, which may have led to inconsistencies between lab sites. Lastly, we did not include patient-reported glucose levels obtained via home monitoring, so it is possible that patients may have experienced hyperglycemia earlier than is reported here.
Additional prospective data is needed to further assess the benefit of prophylactic antihyperglycemics in the prevention of ALP-induced hyperglycemia. There is currently an on-going study in which the safety and efficacy of dapagliflozin in combination with metformin XR will be reviewed in patients with HR + , HER2-, advanced breast cancer while receiving treatment with ALP and fulvestrant (EPIK-B4) [19]. Results will further elucidate prevention tactics for ALP-induced hyperglycemia and will further shape guideline recommendations.

Conclusion
Implementation of a pre-emptive hyperglycemia prevention protocol in patients receiving ALP in a single-center study setting demonstrated fewer grade 3-4 hyperglycemic events compared to rates reported in prior studies. Patients experiencing grade 2 or grade 3 HG had a significantly greater number of risk factors compared to those who did not experience a hyperglycemic event. Optimizing early identification of risk factors and implementing a preventative protocol prior to ALP initiation should be prioritized, as it may lead to fewer dose reductions and discontinuations of therapy due to hyperglycemic events. Additionally, this approach could facilitate the use of ALP in a higher risk patient population that would otherwise not be deemed good candidates for ALP therapy.
Author contributions SB, HM, JF, and KW were responsible for concept and design of the study. SB contributed to collection of the data. EP and H-JL conducted statistical analysis of the data. All authors contributed to analysis and interpretation of the data. SB, with supervision from HM, drafted the manuscript. All authors critically reviewed the manuscript and approved the final version.
Funding Not applicable.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations
Conflict of interest H Moore is an active consultant for Novartis Pharmaceuticals and serves on the EPIK-B4 steering committee and is active PI of EPIK-B4 at Duke University Hospital. S Burnette, HJ Lee, and E Poehlein, J Force, and K Westbrook have no conflicts of interest to report.
Ethical approval and consent to participate All study protocols were reviewed and approved by Duke University Health System's Institutional Review Board.
Consent for publication Not applicable.