Metformin Therapy in Autosomal Dominant Polycystic Kidney Disease: A Feasibility Study

Abstract

Rationale & Objective:

Autosomal dominant polycystic kidney disease (ADPKD) is a common inherited disorder that leads to kidney failure and has few treatment options. Metformin is well tolerated and safe in other patient populations. The primary objective of this clinical trial was to determine the safety and tolerability of metformin in patients with ADPKD and without diabetes mellitus.

Study Design:

Prospective, randomized, controlled, double-blind clinical trial.

Setting & Participants:

N=51 adults 30–60 years of age with ADPKD, without diabetes, and an estimated glomerular filtration rate (eGFR) 50–80 mL/min/1.73 m².

Intervention:

Metformin (maximum dose 2,000 mg/day) or placebo for 12 months.

Outcomes:

Co-primary endpoints were the percentage of participants in each group prescribed at the end of the 12-month period: (a) the full randomized dose, and (b) at least 50% of the randomized dose. Secondary and exploratory outcomes were the effect of metformin compared to placebo on (a) percent change in TKV referenced to height (htTKV in mL/m) and (b) change in eGFR over a 12-month period.

Results:

Mean age was 48 ± 8 years and eGFR was 70 ± 14 mL/min/1.73m². The metformin group had no cases of lactic acidosis and there was one episode of mild hypoglycemia in each group. Participants in the metformin group reported more adverse symptoms, mostly related to gastrointestinal symptoms. 11 of 22 (50%) of metformin-treated participants completed the treatment phase on the full dose compared to 23 of 23 (100%) in the placebo group. 82% of participants on metformin tolerated at least 50% of the dose, compared to 100% in the placebo group. In exploratory analyses, changes in height-adjusted total kidney volume or eGFR were not significantly different between groups.

Limitations:

Short study duration.

Conclusions:

50% or more of the maximal metformin dose was safe and well tolerated over 12 months in patients with ADPKD. Safety of other preparations of metformin as well as its efficacy should be tested in future clinical trials.

Funding:

R21 DK107969–01A1 and Zell Foundation

Trial Registration:

NCT 02903511

PLAIN LANGUAGE SUMMARY

Autosomal dominant polycystic kidney disease (ADPKD) is an inherited disorder that leads to kidney failure. Despite decades of research treatments for ADPKD are limited. Metformin has a track record of safety in patients with type 2 diabetes. Potential benefits of metformin in ADPKD are related to its ability to stimulate AMP-kinase, an important enzyme involved in the regulation of cell metabolism. In this manuscript, we present the results of a pilot clinical trial to assess the safety and feasibility of using metformin in ADPKD over a 12-month period. Metformin was safe, however, only 50% of the participants were able to tolerate the full dose of immediate-release metformin. Future trials should test a better tolerated extended-release formulation for its efficacy to slow the progression of ADPKD.

INTRODUCTION

Autosomal dominant polycystic kidney disease (ADPKD), caused mainly by mutation in either PKD1 or PKD2 encoding polycystin 1 and 2 respectively, is a common familial disorder leading to kidney failure in the majority of affected individuals. It occurs in all ethnicities and races worldwide, with an estimated prevalence of 3.3–4.6 per 10,000 population.^1,2,3 Mean age at initiation of kidney replacement therapy (KRT) is 53–58 years for patients with the more common (80–85%) PKD1 mutation,^4,5 causing significant morbidity and mortality. Despite research progress in ADPKD, the prognosis of patients has not substantially changed over the past 20 years and treatment options are limited.

Exactly how the gene mutations lead to cyst formation and ultimately to kidney failure is unknown, but pathogenetic processes involve cyst epithelial cell proliferation and chloride secretion into the cyst lumen, interstitial inflammatory cell infiltration, and ultimately interstitial fibrosis, tubular atrophy and glomerulosclerosis.⁶ Reduced abundance of functional polycystin 1 or 2 at the cellular level leads to increased signaling of cAMP, which stimulates both proliferation and secretion,^7–9 and increased activity of mammalian target of rapamycin (mTOR), normally repressed by polycystin 1.^10,11 To date, tolvaptan is the only approved intervention to slow kidney disease progression in patients with ADPKD, based on its ability to suppress cAMP signaling and thus cyst growth.¹² However, the use of tolvaptan is constrained by high cost, common side effects, need for continuous laboratory monitoring and the potential for fatal liver injury.

A hallmark of cyst epithelial cells is their altered metabolism, characterized by decreased activity of AMP-activated protein kinase (AMPK) and a shift from mitochondrial respiration towards aerobic glycolysis, similar to cancer cells (Warburg effect).^13–16 Decreased AMPK activity further upregulates mTOR and cell proliferation,^13–16 and also promotes chloride secretion by activating cystic fibrosis conductance transmembrane regulator (CFTR).^17,18 Therefore, drugs that activate AMPK such as metformin, might decrease both cell proliferation and fluid secretion, thus attenuating cyst growth.^19–21 Metformin is inexpensive and has a long safety record for the treatment and prevention of diabetes type 2 and for polycystic ovary syndrome.^22–24 Pharmaco-epidemiological studies have shown that metformin therapy is associated with reduced risk of cardiovascular- and kidney-related endpoints in patients with and without diabetic kidney disease,^22,25–32 which makes it an attractive agent for ADPKD patients.

We herein report the results of the NIDDK funded Feasibility Study of Metformin Therapy in Autosomal Dominant Polycystic Kidney Disease Pilot Trial (NCT 02903511) which tested the safety and tolerability of metformin in ADPKD patients with mildly reduced kidney function (estimated glomerular filtration rate (eGFR) 50–80 mL/min/1.73 m²). A secondary, exploratory goal was to determine the effect of metformin on total kidney volume (TKV) growth and on kidney function decline.

METHODS

Study Design and Population

This pilot trial was a parallel-group, randomized, double-blinded, placebo-controlled trial conducted at the University of Colorado Anschutz Medical Campus. Enrollment occurred between November 2016 and September 2019. Key inclusion criteria were age 30–60 years; diagnosis of ADPKD based on updated Ravine criteria;³³ eGFR 50–80 mL/min/1.73 m²; blood pressure (BP) < 130/80 mmHg; if on antihypertensive medications, stable doses for at least 4 weeks; and subjects had to be free from alcohol or drug dependence and able to provide informed consent. Key exclusion criteria were a diagnosis of diabetes mellitus, current smoking or history of smoking in the past 12 months, history of hospitalization within the last 3 months, severe comorbidities, pregnancy or inability to use contraception for females of childbearing age, and body mass index (BMI) < 21 kg/m².

Description of the Intervention

Participants were randomized in a ratio of 1:1 to either metformin or placebo, administered orally twice a day with meals for 12 months. Subjects were categorized based on a block randomization scheme (i.e., in block of 2 participants) and then a random number generator was used to assign subjects to either metformin or placebo. The starting dose was 500 mg twice a day (bid). If tolerated well, the dose was up-titrated every 2 weeks by 500 mg to a maximal dose of 1000 mg bid. Titration could proceed at longer intervals if bothersome gastrointestinal symptoms occurred. If study participants deemed their gastrointestinal symptoms intolerable at higher doses, the dose was reduced to the previously tolerated dose. If adverse symptoms improved and the participant was willing, a second attempt at dose escalation was made. The criterion for permanent dose reduction was the patient’s inability to continue taking the higher dose. For patients whose eGFR decreased below 45 mL/min/1.73 m² on 2 measurements 2 weeks apart during the study, the maximum dose was reduced to 500 mg bid according to the recommendations of KDIGO (Kidney Disease Improving Global Outcomes).³⁴ If eGFR fell below 30 mL/min/1.73 m² the study medication was to be stopped.

Safety Monitoring, Sample Collection and Measurements

At the baseline visit all participants were given a glucometer, instructed in its use, and asked to check their fasting blood sugars (BS) daily for the first week of treatment, then twice weekly for the following week, and at any time if they experienced a hypoglycemic symptom, and to keep a log of their BS values. The logs were reviewed during scheduled telephone calls. After each dose change of metformin during the titration period participants followed the same BS monitoring schedule. Subjects were also instructed to hold study medication during an intercurrent illness that could precipitate acute kidney injury, such as acute gastroenteritis, febrile illness, or hospitalization. Scheduled telephone visits were conducted every 3 months, during which symptoms, medications, intercurrent illnesses and adverse events were reviewed. Additional telephone visits were performed when needed. Safety laboratories (basic metabolic panel) were obtained at baseline and at 3, 6, 9 and 12 months, to identify a decline in eGFR necessitating metformin or placebo dose reduction. Adherence to the prescribed treatment was assessed by pill counts.

Primary and Secondary Outcomes

The co-primary endpoints were the percentage of participants in each group prescribed at the end of the 12-month period: (a) the full randomized dose, and (b) at least 50% of the randomized dose. The choice of 50% of the prescribed dose is based on previous studies demonstrating that metformin stimulates AMPK activity at low concentration and that ADPKD patients who take less than the full dose may still derive a kidney protective benefit.^35,36 Dose adjustments solely due to low GFR (< 45 mL/min/1.73 m²) were considered as full dose because the exposure to metformin in subjects with reduced metformin excretion is equivalent to the full dose given to subjects with normal GFR. The rationale of having 2 co-primary endpoints was to provide an assessment of the feasibility, in terms of safety and tolerability, to implement either dose in a future Phase 3 trial. We anticipated that not all participants in the metformin arm would complete the study on the full dose (i.e., 2000 mg a day) and that some participants would require a dose reduction due to side effects or intolerance, as metformin is known to cause gastrointestinal adverse events. Hence, a second benchmark was included because participants who require a lower dose due to side effects might still benefit from treatment in a future Phase-3 trial. We considered that the benchmark was achieved if: (1) at least 67% of participants in the metformin group were prescribed the full randomized metformin dose, and (2) at least 80% of participants in the metformin group were prescribed at least 50% of the randomized metformin dose at the end of the12-month intervention period. The rationale for these benchmark cutoffs considered: (1) estimates of tolerability and feasibility used in other biobehavioral trials,³⁷ (2) potential loss to follow-up and dropout rate 15–20% in the metformin group, and (3) participants who take lower than the randomized dose may still benefit from therapy. Secondary and exploratory outcomes evaluated the effect of metformin compared to placebo on (a) percent change in TKV which was referenced to height (htTKV in mL/m) and (b) absolute change in eGFR over the 12-month period. Given the limited sample size and short duration of the trial, the effects of metformin on htTKV and eGFR were considered exploratory only, as defined in the protocol a priori. The direct comparison between the metformin and placebo group was not required to determine if the co-primary endpoints were met. The rationale of having a placebo arm in this safety and feasibility pilot trial was: (i) minimize perception of participant’s side effects; (ii) exploratory evaluation of efficacy endpoints; (iii) examine if pill burden contributed to compliance and (iv) to better assess the safety of metformin. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation,³⁸ based on serum creatinine determinations by isotope dilution mass spectrometry in the central laboratory of Colorado University Hospital. TKV was determined by magnetic resonance imaging (MRI) using a 3.0 T Siemens system performed at baseline and at study end, using methods established by the Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP).³⁹ To ensure consistency, volumetric measurements of TKV were done on de-identified images by 1 reader (Dr. Wei Wang) who was blinded to study assignment, in one batch after study completion, using Analyze software (Analyze 11.0, Mayo Foundation, Rochester, MN). Disease severity was categorized according to the Mayo Imaging Classification.⁴⁰

Study Oversight

An independent Data Safety Monitoring Board convened twice during the study and did not identify significant safety issues. The study was approved by the Colorado Multiple Institutional Review Board (COMIRB) under protocol number 16–0802, and all subjects provided their informed written consent. The study was registered at clinicaltrials.gov (NCT 02903511) and an investigational new drug application was filed with the Food and Drug Administration who deemed the study exempt.

Statistical Analyses, Sample Size and Power Calculations

Participant baseline characteristics were summarized using means and standard deviations, medians and interquartile ranges, or frequencies and proportions, as appropriate for all enrolled participants and, separately, by randomized group. Standardized differences were calculated and reported across treatment groups. We tabulated the proportions of the participants assigned to each of the 2 treatment groups who at the end of the study: (a) were prescribed the full randomized metformin dose (allowing for dose adjustments for decreased GFR), and (b) were prescribed at least 50% of the randomized dose. Percentage of subjects who experienced various types of adverse events, including hospitalizations, were tabulated between treatment groups. A two-sample t-test was used to compare the change in htTKV and eGFR over 12-months between the metformin and placebo groups. A two-way interaction between treatment effect and baseline htTKV and Mayo Class status was assessed using likelihood-ratio tests.

The randomization goal for this pilot trial was 50 participants, with 25 individuals each randomized to metformin and placebo. The statistical power proposed for attaining the 67% benchmark was at least 0.80 in the metformin group if the true percent prescribed the full randomized metformin dose was at least 90%. The statistical power for attaining the 80% benchmark was at least 0.80 in the metformin group if the true percent prescribed at least 50% of the full randomized metformin dose was at least 97%. A one-sided type I error of 0.025 was used in power analysis because of co-primary outcomes. An exact test was used to compare the observed proportion of participants who completed the study on the full dose or on ≥ 50% of the full dose with the proposed benchmarks. A significance level of one-sided 0.025 was used for the co-primary endpoints and two-sided 0.05 for the exploratory endpoints. Statistical analyses were performed with SAS version 9.4 (SAS Institute).

RESULTS

Of the 139 subjects who were prescreened for participation in this trial, 51 participants aged 30–60 years with a diagnosis of ADPKD and an eGFR of 50–80 mL/min/1.73 m² were randomized over a period of 35 months to either metformin (n=26) or placebo (n=25; Figure 1). One participant in the metformin group withdrew prior to study start. A total of 3 participants in the metformin and 2 in the placebo group were lost to follow-up prior to the final study visit at 12 months.

Figure 1.

CONSORT diagram: Flow of study participants from screening to study completion.

Table 1 shows the baseline characteristics of all randomized participants and by treatment assignment. Mean (SD) age was 48±8 years, 37% were male, and 98% were Caucasian. Hypertension was present in 84%, and 65% were taking a renin-angiotensin-aldosterone system (RAAS) blocking drug. Mean (SD) eGFR was 70±13 mL/min/1.73 m². Sixty-nine percent of participants were in the higher risk Mayo Imaging Classes 1C, 1D or 1E. Demographic and most clinical characteristics were distributed similarly across randomized groups. Participants in the metformin arm had a higher prevalence of hypertension, and had larger TKV than participants in the placebo arm.

Table 1.

Baseline Characteristics of Randomized Participants Overall and by Treatment Assignment

Variable	Total (n=51)	Metformin (n=26)	Placebo (n=25)	Standardized Difference
Age, years	48±8	48±8	48±7	0.10
Age at diagnosis, y	29±12	30±12	29±13	0.11
Sex, % Male, No. (%)	19 (37)	11 (42)	8 (32)	0.21
Race/Ethnicity, Non-Hispanic White, No. (%)	50 (98)	26 (100)	24 (96)	0.29
Hypertension, No. (%)	43 (84)	24 (92)	19 (76)	0.46
Weight, kg	85±18	87±21	84±15	0.16
BMI, kg/m2	28.8±5.6	29.2±6.7	28.4±4.3	0.15
SBP, mmHg	125±12	124±12	125±12	0.07
DBP, mmHg	81±8	80±8	82±9	0.18
CKD-EPI eGFR, mL/min/1.73m2	70±13	68±13	72±13	0.37
eGFR < 60 mL/min/1.73m2, No. (%)	11 (22)	6 (23)	5 (20)	0.08
Total Kidney Volume (TKV), mL [IQR]	1229 [960–2299]	2101 [1015–2622]	1156 [877–1760]	0.60
Height adjusted TKV, mL/m [IQR]	715 [533–1353]	1281 [587–1580]	688 [517–1004]	0.61
ACEi/ARB, No. (%)	33 (65)	20 (77)	13 (52)	0.54
Mayo Class				1.11
Mayo Class 1A, No. (%)	3 (6)	1 (4)	2 (8)
Mayo Class 1B, No. (%)	13 (25)	6 (23)	7 (28)
Mayo Class 1C, No. (%)	22 (43)	8 (31)	14 (56)
Mayo Class 1D, No. (%)	8 (16)	7 (27)	1 (4)
Mayo Class 1E, No. (%)	5 (10)	4 (15)	1 (4)

Data are mean ± S.D. or median [interquartile range]. BMI, body-mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure, CKD-EPI eGFR; estimated glomerular filtration rate by the Chronic Kidney Disease Epidemiology Collaboration equation; ACEi, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker.

Note: The standardized difference was calculated to measure the distance between two groups in a variable. There is only one single value of the standardized difference for the Mayo Class, because standardized difference is defined for a variable and cannot be calculated for each category of a categorical variable with multiple categories.

Primary Endpoint Results

Among participants randomized to metformin who completed the study, 11 of 22 (50%) completed the treatment phase on the full dose, including 2 participants who required (per protocol) a dose reduction to 1000 mg daily for safety after their eGFR dropped below 45 mL/min/1.73 m²; they did not report significant side effects. Seven additional participants completed the treatment phase on at least 50% of the full metformin dose. Hence, at least 80% of participants (18 of 22) tolerated at least 50% of the prescribed dose (Table 2). The proportion of participants that completed the study on the full randomized metformin dose was not significantly higher than the prespecified benchmark of 67% (p>0.99) and the proportion of participants that completed the study on ≥ 50% of the full randomized dose was not significantly higher than the prespecified benchmark of 80% (p=0.42).

Table 2.

Co-Primary Endpoints

Co-primary Endpoints	Metformin (n=22)	Placebo (n=23)
Completing on Full Dose (Includes dose reduction due to eGFR decline)	11 (50%)	23 (100%)
Completing on ≥ 50% of the Full Dose	18 (82%)	23 (100%)

Of the 7 participants on reduced dose metformin, 4 were taking 1500 mg daily and 3 were taking 1000 mg daily at 12 months. Two did not tolerate more than 500 mg daily and 2 stopped metformin altogether because of side effects but came back for the end-of-study visit, thus completed the study. Over the course of the treatment phase, the median (IQR) dose of metformin prescribed was 1500 (1000–2000) mg daily. The most common reason for reducing the metformin dose were gastrointestinal complaints. Among participants in the placebo group, 23 of 23 (100%) completed the treatment phase on the full dose (Table 2). Overall adherence by pill count was 96% for the metformin group and 97% for the placebo group. Of note, changes in blood pressure, weight, body mass index and serum glucose were not different between the treatment groups (Supplemental Table 1).

Safety

Participants in the metformin group reported more adverse symptoms, mostly related to the gastrointestinal tract, which resolved either spontaneously or after dose reduction of metformin (Table 3). Gastrointestinal symptoms were also common in the placebo group but did not require dose modification because participants reported these symptoms as tolerable. Two safety events required hospitalization in the metformin group; both were considered unrelated to study drug. One participant was hospitalized for acute pyelonephritis which was treated successfully with antibiotics; the other participant was hospitalized for observation of multiple symptoms with no pathologic findings on laboratory and imaging studies, ultimately diagnosed as a viral infection. In both cases metformin was held during hospitalization and resumed afterwards without further problems. Two mild hypoglycemia episodes were reported by patients, one in each group. Both patients experienced lightheadedness; blood glucose level was 49 mg/dL in the participant on metformin and 54 mg/dL in the participant assigned to placebo. The patient on metformin had been fasting all day and was counseled to eat before exercising and taking study drug, and no further hypoglycemic symptoms occurred. The other participant on placebo reported the episode at the 9-month visit, and no further symptoms occurred without any dosing changes.

Table 3.

Effect of Metformin on Hospitalizations and Adverse Events (AE)

Event	Metformin n=26			Placebo n=25
	Event No.	Crude Rate, %	Rate per patient-yr	Event No.	Crude rate, %	Rate per patient-yr
Hospitalizations	2	8	0.08	0	0	0
Hypoglycemia	1	4	0.04	1	4	0.04
Nausea	11	42	0.42	4	16	0.16
Vomiting	7	27	0.27	3	12	0,12
Diarrhea	16	62	0.62	7	28	0.28
Acid reflux	2	8	0.08	4	16	0.16
Flatulence	4	15	0.15	0	0	0
Lightheadedness	6	23	0.23	4	16	0.16
Headaches	3	12	0.12	4	16	0.16
Constipation	1	4	0.04	0	0	0
Abdominal cramping/pain	2	8	0.08	3	12	0.12
Weakness	5	19	0.19	0	0	0
Viral illness	5	19	0.19	7	28	0.28
Urinary tract infections	1	4	0.04	2	8	0.08
Sinus infections	5	19	0.19	2	8	0.08

Exploratory Secondary Endpoints:

Percent increase in height-adjusted TKV (htTKV) and absolute eGFR decline over 12 months were pre-specified exploratory secondary endpoints. For one participant in each group, it was difficult to accurately determine htTKV on the follow-up images due to quality, therefore these 2 participants were excluded from the secondary analyses. Among 43 participants with evaluable htTKV at baseline and at 12 months, 21 were in the metformin group and 22 in the placebo group (Table 4). Changes in htTKV and eGFR were not significantly different between the groups, but numerically the decline in eGFR was much smaller in the metformin group (−0.41±1.81 vs −3.35±1.70 mL/min/1.73 m², p=0.24) although these patients had significantly larger kidneys and more hypertension (i.e. more severe disease) than those in the placebo group, thus were expected to have a larger eGFR decline. To explore further any potential signal for efficacy, we repeated these analyses for participants with htTKV > 600 mL/m, for those with htTKV > 800 mL/m and for those in the more severe Mayo Imaging Classes 1C–E. Changes in htTKV were numerically smaller in all 3 subgroups assigned to metformin compared to placebo, and this was statistically significant in the subgroup with htTKV > 800 mL/m (0.81±1.92 % vs 5.86±0.86 %, p=0.03) (Table 4). In supplemental table 2 and 3 we show baseline characteristics and effects of metformin on the exploratory endpoints by participants that completed the study on the full dose compared to those who completed on at least 50% of the full dose.

Table 4.

Exploratory Outcomes of Annual Changes in Height-Adjusted Total Kidney Volume and Estimated Glomerular Filtration Rate

	N	Metformin Mean (SE)	Placebo Mean (SE)	p-Value
htTKV (full cohort; % change)	43 (21 M; 22 P)	3.45 (1.30)	3.15 (1.61)	0.88
eGFR (mL/min/1.73m2)	43 (21 M; 22 P)	−0.41 (1.81)	−3.35 (1.70)	0.24
Participants with baseline TKV > 600 ml/m					Treatment*Baseline htTKV Interaction
htTKV (% Change)	25 (13 M; 12 P)	1.79 (1.75)	4.73 (1.86)	0.26	0.35
eGFR (mL/min/1.73m2)	25 (13 M; 12 P)	−1.29 (2.44)	−4.17 (1.96)	0.38	0.83
Participants with baseline TKV > 800 ml/m					Treatment*Baseline htTKV Interaction
htTKV (% Change)	18 (11 M; 7 P)	0.81 (1.92)	5.86 (0.86)	0.03	0.22
eGFR (mL/min/1.73m2)	18 (11M; 7 P)	−3.00 (2.57)	−3.86 (3.22)	0.84	0.24
Participants in Mayo Imaging Classes 1C, 1D and 1E					Treatment*Baseline Mayo Class Interaction
htTKV (% Change)	26 (14 M; 12 P)	1.92 (1.62)	4.39 (2.43)	0.41	0.76
eGFR (mL/min/1.73m2)	26 (14 M; 12 P)	−1.57 (2.50)	− 1.85 (2.35)	0.94	0.52

M= metformin; P= placebo. P-values are based on the two-sample t-test comparing changes in parameters between the 2 groups

DISCUSSION

Prevention of ADPKD progression remains a high priority. Available therapies to slow the loss of kidney function in this patient population are limited, therefore additional clinical trials to evaluate promising treatments are needed. To inform the design of a large-scale clinical trial of metformin therapy for ADPKD, we conducted this pilot trial to assess safety, tolerability and adherence to metformin 1000 mg twice a day. We found that the feasibility of metformin was only observed at an absolute level, as 82% of participants assigned to metformin completed the study on at least 50% of the dose, however, these findings were not statistically significant. Only 50% of participants were able to tolerate the full metformin dose at the end of the intervention period as the other 50% required some degree of dose reduction due to gastrointestinal symptoms. There was no evidence that a dose of 2000 mg/day of the immediate-release metformin formulation is feasible in a long-term Phase 3 clinical trial. Metformin was safe during the course of the study. There were no significant episodes of hypoglycemia and no case of lactic acidosis; and adherence was > 95% in each group during the study.

The rate of gastrointestinal symptoms observed in our pilot trial is consistent with the reported side effect profile of metformin when used for the treatment of type 2 diabetes.^41,42 Diarrhea is the most frequent adverse symptom, occurring in 8–62% of exposed individuals and leading to discontinuation in approximately 5% of trial participants.^41,42 Nausea is also frequently observed in metformin trials for type 2 diabetes, occurring in 8–12% of participants.⁴² Diarrhea and nausea were reported by both patient groups in our trial, but were more frequent and more severe in the metformin group, leading to dose reduction or withdrawal only in this group. Although dose reduction often improves gastrointestinal symptoms, approximately 5% of people cannot tolerate metformin at all.⁴¹ Underlying mechanisms for gastrointestinal toxicity are not entirely clear. Direct effects of metformin on enterocytes, changes in the gut microbiome, inhibition of bile acid reabsorption, and genetic factors have been implicated.^41,42 Extended-release (XR) metformin formulations reportedly cause fewer gastrointestinal side effects and are associated with better adherence.⁴³ These formulations should be tested in future metformin trials in patients with ADPKD.

Many experimental studies support the renoprotective potential of metformin in ADPKD and in non-ADPKD chronic kidney disease, based on its anti-inflammatory and antifibrotic effects and its ability to improve vascular function.^{20,21,43–48}. Metformin stimulates AMPK activity at low concentrations³⁵ suggesting that lower doses may be sufficient for cyst inhibiting and renoprotective effects. Although our study was not powered to detect significant differences in changes of htTKV or eGFR between the groups, there was a signal that the annual percent increase in htTKV and decline in eGFR were smaller in the metformin group, and similarly for participants completing the study on full as compared to reduced dose. This occurred contrary to the expectation of a larger increase in htTKV and faster eGFR decline in participants with much larger baseline htTKV.

This is one of the first completed randomized, double-blind, placebo-controlled trial of metformin therapy for ADPKD. Previously we had performed a retrospective database analysis using the Intermountain Integrated Health Care database. We compared the course of 31 adult ADPKD patients with diabetes type 2 treated with metformin with 31 matched diabetic ADPKD patients with similar baseline kidney function (mean eGFR 48±13 mL/min/1.73 m²) who were treated with other antidiabetic drugs. After a median follow-up of 4.5 years, ESKD occurred in 16% of metformin-treated compared to 29% of other-drug-treated patients, which was statistically significant (p=0.02). No information on changes in TKV was available in this population.⁴⁹

Another retrospective study compared 7 diabetic adults with ADPKD and stage 3 CKD (mean eGFR 48±11 mL/min/1.73 m² at baseline) who were treated with metformin at least 1000 mg/day for 3 years, with 7 nondiabetic subjects with ADPKD, matched for sex, age, and baseline eGFR.³⁶ Loss of eGFR, estimated by a linear mixed model, was 0.9 mL/min/1.73 m² per year in the metformin-treated patients, significantly less than the loss of 5 mL/min/1.73 m² per year observed in the matched control patients (p=0.002).

A second randomized trial of metformin therapy for ADPKD, conducted at 2 clinical centers (Tufts Medical Center and University of Maryland) in the United States, has recently been completed (NCT 02656017). It was also designed as a double-blind tolerability and feasibility study in 97 non-diabetic adult patients with ADPKD and normal or near-normal renal function, with 26 months of follow-up.⁵⁰ Mean baseline eGFR is 87±19 mL/min/1.73 m² (vs 70±13 in our study), median baseline htTKV is 610 mL/m (vs 742 mL/m in our study) and 48% of participants are in higher-risk Mayo Imaging Classes (1C, 1D and 1E), as compared to 69% in our study. Hence, disease severity is less than in our pilot trial.

A unique strength of our pilot trial is its objective to assess the safety, tolerability, and adherence to metformin specifically in ADPKD patients with mildly reduced renal function. The small sample size, 10% attrition, short duration of the study and imbalance in htTKV among treatment groups preclude definitive conclusions, but we observed a signal of slowed eGFR decline by 2.9 mL/min/1.73m².

In conclusion, metformin 500–1000 mg twice daily had a favorable adherence and safety profile over 12 months in participants with ADPKD with a mean (SD) eGFR of 70±13 mL/min per 1.73 m². Because of the therapeutic potential of metformin for ADPKD, further large-scale trials are warranted, using extended-release formulations and randomizing patients after a run-in period to exclude intolerant participants.

Data Sharing:

Data will be publicly available through the data repository of the NIDDK, National Institutes of Health (NIH). Data from the primary outcome results will be submitted to the repository within 6 months of the date of publication.