Use of Metformin in the Setting of Mild-To-Moderate Renal Insufficiency

Reviews/Commentaries/ADA Statements REVIEW

Use of Metformin in the Setting of Mild-to-Moderate Renal Insufﬁciency

1 KASIA J. LIPSKA, MD hepatic gluconeogenesis without raising 2 CLIFFORD J. BAILEY, PHD, FRCP fi 3 insulin levels, it rarely leads to signi cant SILVIO E. INZUCCHI, MD hypoglycemia when used as a monother- apy (8,11). As a result, metformin is widely considered an ideal first-line agent for the treatment of type 2 diabetes, as common clinical conundrum faces despite multiple trials of intensive glu- recommended by several clinical guide- all U.S. practitioners treating pa- cose control using a variety of glucose- lines (12–14). A fi tients with type 2 diabetes. Today’s lowering strategies, there is a paucity of In addition to such bene ts, metfor- U.S. Food and Drug Administration pre- data to support specificadvantageswith min reduces the risk of developing di- scribing guidelines for metformin contra- other agents on cardiovascular outcomes abetes in individuals at high risk for the indicate its use in men and women with (5–7). disease (15) and has been considered as a serum creatinine concentrations $1.5 In the original UK Prospective Di- reasonable “off-label” approach in se- and $1.4 mg/dL ($132 and $123 abetes Study (UKPDS), 342 overweight lected individuals for diabetes prevention mmol/L), respectively. In a patient toler- patients with newly diagnosed diabetes (16). ating and controlled with this medication, were randomly assigned to metformin should it automatically be discontinued therapy (8). After a median period of 10 — as the creatinine rises beyond these cut years, this subgroup experienced a 39% HISTORICAL PERSPECTIVE De- fi points over time? Stopping metformin of- (P = 0.010) risk reduction for myocardial spite these proven bene ts, metformin ten results in poorly controlled glycemia infarction and a 36% reduction for total remains contraindicated in a large seg- and/or the need for other agents with their mortality (P = 0.011) compared with con- ment of the type 2 diabetic population, own adverse-effect profiles. Moreover, is ventional diet treatment. Similar benefits largely because of concerns over the rare the now widespread use of estimated glo- were not observed in those randomly as- adverse effect of lactic acidosis. For these merular filtration rate (eGFR) in lieu of signed to sulfonylurea or insulin. In an reasons, the drug has been restricted to serum creatinine levels creating even 8.5-year posttrial monitoring study, after individuals with normal creatinine levels more confusion, especially in those with participants no longer were randomly as- as a surrogate for renal competence. abnormalities in one but not the other in- signed to their treatments, individuals Other contraindications (e.g., any signif- direct measure of renal function? Indeed, originally assigned to metformin (n = icant hypoxemia, alcoholism, cirrhosis, a more than a decade and a half after met- 279) continued to demonstrate a reduced recent radiocontrast study) also increase formin became available in the U.S., de- risk for both myocardial infarction (rela- the risk for or the consequences of lactic bate continues over the best approach in tive risk 33%, P = 0.005) and total mor- acidosis, but these are not the topic of this these settings (1–3). How many patients tality (relative risk 27%, P = 0.002) (9). review. are unable to receive this medication on The latter results are even more impres- Metformin belongs to the biguanide drug class (previous members include the basis of guidelines which, although sive when one considers that HbA1c levels well intentioned, are somewhat arbitrary in all initially randomly assigned groups phenformin and buformin), developed and outdated based on modern assess- had converged within 1 year of follow-up. for lowering glucose in the 1950s. Initial ments of renal status? Unlike sulfonylureas, thiazolidine- enthusiasm for biguanides was tempered diones, and insulin, metformin is weight over the next two decades by the growing ADVANTAGES OF neutral (10), which makes it an attractive recognition of their risk of lactic acidosis. METFORMIN—Thereissomeevi- choice for obese patients. Furthermore, A marked reduction in biguanide use dence that early treatment with metformin the management of type 2 diabetes can occurred in the mid-1970s because phen- is associated with reduced cardiovascular be complicated by hypoglycemia, which formin, extensively adopted in clinical morbidity and total mortality in newly can seriously limit the pursuit of glycemic practice, was implicated in a number of diagnosed type 2 diabetic patients (4). control. Here, too, metformin has advan- fatal cases of this severe metabolic de- However, the data come from a small sub- tages over insulin and some types of in- compensation (17). The association with group of a much larger trial. In contrast, sulin secretagogues; by decreasing excess lactic acidosis eventually led to its with- drawal from the market. Importantly, lac- ccccccccccccccccccccccccccccccccccccccccccccccccc tic acidosis with phenformin seems to occur ~10–20 times more frequently From the 1Robert Wood Johnson Clinical Scholars Program, Yale University School of Medicine, New Haven, Connecticut; the 2Department of Life and Health Sciences, Aston University, Birmingham, U.K.; and the than with metformin (18). In contrast to 3Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut. metformin, modestly raised phenformin Corresponding author: Silvio E. Inzucchi, [email protected]. concentrations may reduce peripheral Received 15 December 2010 and accepted 16 March 2011. glucose oxidation and enhance peripheral DOI: 10.2337/dc10-2361 © 2011 by the American Diabetes Association. Readers may use this article as long as the work is properly lactate production, which can increase cir- cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/ culating lactate levels. In fact, phenformin licenses/by-nc-nd/3.0/ for details. levels correlate with lactate concentration, care.diabetesjournals.org DIABETES CARE, VOLUME 34, JUNE 2011 1431 Metformin in renal insufficiency whereas metformin levels do not (19). In updated clinical practice guidelines is- intestine and salivary glands as well as addition, ~10% of European Caucasians sued by the American Diabetes Associa- in the kidney (25). It has a plasma elimina- have an inherent defect in phenformin hy- tion, for example, do not actually discuss tion half-life of 6.2 h and is renally elimi- droxylation, which may lead to drug accu- renal thresholds (16) but refer the reader nated both by filtration and active tubular mulation and, as a result, elevated lactate to a separate consensus statement for pre- secretion (26). levels (20). scribing details. This statement, authored In careful experiments, Tucker et al. The experience with phenformin re- by members of the American Diabetes As- (27) studied metformin kinetics in 4 sulted in cautious use of metformin in sociation and European Association for healthy subjects and 12 type 2 diabetic Europe. In the 1980s, the creatinine cut the Study of Diabetes, reports that met- subjects and found that plasma renal points for contraindication to metformin formin seems safe unless eGFR falls to clearance of metformin highly correlated were considered to be appropriate at 1.4 ,30 mL/min per 1.73 m2 (12). with creatinine clearance (CrCl; r = 0.85, mg/dL in women and 1.5 mg/dL in men. Clinical guidelines outside of the U.S. P , 0.001). However, the relationship This was based on the calculated ability to incorporate the eGFR for determination between physiological clearance of an ac- remove 3 g of metformin (an amount of metformin safety. In the U.K., for tual oral dose and CrCl was much weaker slightly beyond the maximum daily U.S. example, prescribing guidelines consider (r = 0.66, P , 0.01). Therefore, the inves- dose) at steady-state levels within 24–48 h. both creatinine and eGFR for assessing tigators postulated that other factors may In fact, the ability to comfortably remove treatment eligibility. The National Insti- impact this relationship, such as perhaps the drug extends up to creatinine levels of tute for Health and Clinical Excellence gastrointestinal absorption of metformin 1.8–2.0 mg/dL, but the cut points chosen recommends reviewing the clinical cir- in patients with renal failure and/or non- were intentionally set lower to ensure that cumstances when serum creatinine ex- renal clearance of a small amount of the those patients who may be lost to follow- ceeds 130 mmol/L (1.5 mg/dL) or eGFR drug. up and whose creatinine levels increase falls below 45 mL/min per 1.73 m2.The In another pharmacokinetic study (28), over time would not be at risk for appre- National Institute for Health and Clinical asingle850-mgdoseofmetforminwas ciable drug accumulation. Excellence further specifies that metfor- given to 21 healthy subjects and 13 subjects Metformin was not approved in the min be stopped if serum creatinine ex- with renal insufficiency (mild to severe). U.S. until December of 1994 and was ceeds 150 mmol/L (1.7 mg/dL) (a higher In the control group (data presented are marketed in 1995. The experience with threshold than in the U.S.) or eGFR is be- mean 6 SD) (mean CrCl 112 6 8mL/min), phenformin led to judicious labeling and low 30 mL/min per 1.73 m2 (14). In con- average renal metformin clearance was recommendations for careful monitoring trast, the Canadian Diabetes Association 636 6 84 mL/min, whereas in mild chronic of adverse events. Its new-drug applica- practice guidelines are now based solely kidney disease (CKD) (CrCl 61–90 mL/min; tion was the subject of extensive review. on eGFR, recommending caution with mean 73 6 7) clearance was reduced at The risk for lactic acidosis was estimated eGFR ,60 mL/min per 1.73 m2 and con- 384 6 122 mL/min. The mean renal clear- to be ~3 cases per 100,000 patient-years traindicating its use with eGFR ,30 ance of metformin was lower in subjects based on cases reported from France, the mL/min per 1.73 m2 (23). The Australian with moderate (CrCl 31–60 mL/min; mean U.K., and other countries where pharma- Diabetes Society practice guidelines simi- 41 6 9) and severe (CrCl 10–30 mL/min; covigilance information was available (21). larly recommend against metformin mean 22 6 6) CKD, measuring 108 6 57 Similar estimates were quoted from Sweden with eGFR ,30 mL/min per 1.73 m2 and 130 6 90 mL/min, respectively. Sim- (2.4 cases per 100,000 patient-years), and caution with eGFR 30–45 mL/min ilarly, maximum concentration and the where the number of cases appeared to per 1.73 m2 (24). Thus, although there area under the concentration time curve be decreasing despite rising clinical use is clear recognition that renal failure may were increased in individuals with mod- of metformin. After careful deliberation, be a risk factor for adverse events with erate to severe CKD compared with those metformin was approved by the U.S. metformin use, there is significant diver- with mild CKD or normal renal function. Food and Drug Administration, with lactic gence in opinion across the globe regard- Based on the regression analysis, both CrCl acidosis risk seen as small, although inev- ing the optimal definition of safety. and age were found to be important pre- itable with future widespread availability dictors of metformin clearance. This study of the drug. METFORMIN did not provide evidence for specific PHARMACOKINETICS—The prin- thresholds at which lactate production CLINICAL PRACTICE cipal reason for carefully setting renal may begin to rise. GUIDELINES—The prescribing infor- thresholds is that metformin is eliminated These reports have relied on in- mation for metformin in the current label unchanged primarily by the kidneys. formation derived from single doses of is explicit with respect to renal contra- Thus, one of the most important risk metformin, which may not reflect chronic- indications, based on serum creatinine factors for elevated metformin concentra- treatment pharmacokinetics. In contrast, cut points. It proscribes use at or above tions (which are proposed to lead to lactic few reports have assessed the impact of the 1.4 and 1.5 mg/dL levels in women acidosis) is the inability to clear the drug renal insufficiency on metformin clear- and men, respectively. The recently up- efficiently. Metformin has a 50–60% bio- ance during long-term use. Indeed, one dated Kidney Disease Outcomes Quality availability and is absorbed mainly in the such study (29) concluded that metformin Initiative guidelines from the National small intestine. It does not appear to bind can be efficiently cleared in mild-to- Kidney Foundation are perfectly consis- appreciably to plasma proteins. The max- moderate CKD. In this investigation, 24 tent with the label (22). Yet, certain U.S. imum plasma concentration is observed older patients (aged 70–88 years) were practice guidelines substantially differ in ~2 h after oral dosing, typically reaching a administered metformin 850 mg/day or – m m – their recommendations for metformin Cmax of 1 2 g/mL (~10 mol/L). Metformin 1,700 mg/day based on CrCl of 30 60 use related to renal status. The annually accumulates in the walls of the small mL/min (n = 11) or .60 mL/min (n = 13),

1432 DIABETES CARE, VOLUME 34, JUNE 2011 care.diabetesjournals.org Lipska, Bailey, and Inzucchi respectively. After 2 months, metformin from the Cockroft-Gault and eGFR from Salpeter et al. (36) pooled data from 347 remained in the therapeutic range and lac- the Modification of Diet in Renal Disease comparative trials and cohort studies. tate within the reference limits in all par- equation) have been developed to incor- Not a single case of lactic acidosis was ticipants. In addition, the measured levels porate known demographic and clinical found in .70,000 metformin patient- of metformin and lactate were not statisti- factors affecting serum concentrations. years or .55,000 nonmetformin person- cally different between those with and These equations have their own inherent years. In this analysis, 53% of prospective without renal impairment (29). shortcomings, such as residual limita- studies allowed for inclusion of renal insuf- Another recent study (30) evaluated tions with respect to age and race, un- ficiency, but patient-level serum creatinine metformin levels in patients with type 2 derestimation of GFR in the context of concentrations were not available for re- diabetes and varying renal function. GFR diabetic renal disease (Cockroft-Gault view. Based on statistical inference, the es- was estimated based on cystatin C levels. and Modification of Diet in Renal Disease) timated upper limit of true incidence was The median dose of metformin was 1,500 (31), and in obese individuals (Modifica- 4.3 and 5.4 cases per 100,000 patient-years mg/day. The median serum level of met- tion of Diet in Renal Disease) (32). How- in the metformin and nonmetformin formin was 4.5 mmol/L (~0.6 mg/mL) ever, they provide better estimation of groups, respectively. This investigation (range 0.1–20.7) in patients with eGFR renal function than creatinine alone. suggests that lactic acidosis is extremely .60 mL/min per 1.73 m2 (n = 107), 7.7 Moreover, development of new estimat- rare and the incidence does not differ in mmol/L (~1.0 mg/mL) (0.1–15.2) with ing equations, such as the Chronic Kidney those treated with metformin versus other eGFR 30–60 mL/min per 1.73 m2 (n = Disease Epidemiology Collaboration agents. 21), and 8.9 mmol/L (~1.1 mg/mL) (6.0– equation, may allow for even more accu- In a large, nested, case-control anal- 18.6) with eGFR ,30 mL/min per 1.73 rate estimates of renal function in the fu- ysis of the U.K. general practice research m2 (n = 9). Notably, there were wide var- ture. Finally, dosing considerations by the database (11), the crude incidence of lac- iations in these levels within each group, Food and Drug Administration for other tic acidosis was even lower at 3.3 cases per with few patients having serum levels medications (e.g., sitagliptin, fenofibrate) 100,000 person-years among metformin .20 mmol/L (.~2.6 mg/mL). However, are generally based on CrCl estimated users and 4.8 cases per 100,000 person- the “unsafe” metformin concentration is from such calculations and not on creati- years among sulfonylurea users (in very not really known. At usual clinical doses nine levels themselves. close agreement to the estimates of 3 and schedules, steady-state plasma con- and 2.4 cases per 100,000 patient-years centrations are generally ,1 mg/mL (,7.8 LACTIC ACIDOSIS from Europe and Scandinavia before met- mmol/L). Maximum plasma levels during ASSOCIATED WITH formin’s U.S. approval). Given all of these controlled clinical trials do not generally METFORMIN THERAPY—Even findings, some (37) have argued that the exceed 5 mg/mL (38.8 mmol/L), but these though elevated metformin concentra- occurrence of lactic acidosis with metfor- have not typically enrolled CKD patients. tions have been proposed to lead to lactic min use is merely coincidental and that Moreover, whether measurement of met- acidosis, there are few data regarding the there is no tangible evidence from pro- formin levels actually can aid in the pre- level predisposing to hyperlactatemia. In spective observational studies or clinical diction of lactic acidosis risk remains fact, multiple studies suggest that elevated trials that the drug increases its incidence. unclear. Therefore, although these stud- circulating lactate levels, often attributed Of course, all these data have been col- ies provide some information on the re- to metformin, may actually not be caused lected in the context of contemporaneous lationship between renal function and by the drug. First, lactic acidosis occurs in strict metformin-prescribing guidelines. metformin concentrations, they do not patients with type 2 diabetes more fre- It is possible that looser restrictions may clarify the issue of toxicity and lactic quently than in the general population; in have led to more frequent occurrence of acidosis risk. some reports, the observed rate appears to lactic acidosis. Many of the early pharmacokinetic be similar in patients on metformin versus In summary, lactic acidosis remains studies with metformin actually relied on other glucose-lowering agents (11). Sec- exceedingly rare in clinical trials and co- CrCl based on 24-h urine collection for ond, metformin and lactate levels do not hort studies of metformin therapy. More- creatinine. How well the current serum necessarily appear to correlate, such that over, the available data suggest that lactate creatinine cut points reflect the ability to higher metformin concentrations do not levels and risk of lactic acidosis do not effectively clear the drug also is unknown. consistently occur in those with more se- differ appreciably in patients taking this Creatinine levels, in general, vary in- vere degrees of lactic acidosis (33,34). Fi- versus other glucose-lowering agents. versely with GFR. However, important nally, metformin levels are not linked to Thus, the long-proclaimed causal rela- limitations to the estimation of renal mortality in those who develop lactic ac- tionship between metformin and lactic function with creatinine should be con- idosis, perhaps reflecting the primary ef- acidosis remains in question. sidered. First, serum creatinine can only fect of the underlying cause of the acidosis be used reliably in patients with stable (e.g., hypoxia, hemodynamic compro- CURRENT USE OF kidney function. Second, variation in mise) on outcomes rather than incrimi- METFORMIN IN CKD—Given the creatinine production may differ among nating metformin itself (33–35). current contraindications in the U.S., and within individuals over time, espe- Although lactic acidosis remains a some might consider it a challenge to cially if there are significant changes in recognized, albeit rare, adverse event conduct a new clinical trial to evaluate the muscle mass or in physical activity. Var- associated with metformin, the number use of metformin in individuals with iability in creatinine secretion, extrarenal of lactic acidosis cases continues to be various degrees of impaired renal func- creatinine excretion, assay method, and very small, particularly when one consid- tion, taking into account new criteria for equipment can all affect serum measure- ers the widespread use of this drug. In the assessing glomerular filtration. Yet, evi- ments. Calculated estimates (clearance largest updated Cochrane meta-analysis, dence suggests that metformin is often care.diabetesjournals.org DIABETES CARE, VOLUME 34, JUNE 2011 1433 Metformin in renal insufficiency already used in practice outside of the cut points in the context of metformin confirmation of the mortality benefit current labeling contraindications, pre- therapy. In a review (44) of prescribing in similar patient cohorts in prospective scribed in full knowledge of the relevant practices in the U.K., appropriate use of trials. cutoffs (38–41). For example, in a review the drug was defined on the basis of cre- Although these data are reassuring, (41) of restrictions to metformin therapy atinine level #1.7 mg/dL. Of 11,297 pa- we must note that there are no random- conducted in Scotland, 24.5% of metfor- tients meeting those criteria, 82% had an ized clinical trials that specifically evalu- min users had filled a prescription despite eGFR ,90, 25.5% ,60, and 2.8% ,30 ated the safety of metformin use and active contraindications (3.4% had the mL/min per 1.73 m2. The authors cal- potential cardiovascular benefits in pa- specific local exclusion of a serum creati- culated that the eGFR threshold of 36 tients with CKD. nine $1.7 mg/dL recorded twice on dif- mL/min would result in a similar number ferent days within 4 weeks). A single case of patients becoming ineligible for met- PROGRESSION OF CKD IN of lactic acidosis during 4,600 patient- formin compared with the serum creati- PATIENTS WITH DIABETES—Re- years of follow-up occurred in a patient nine threshold of 1.7 mg/dL (although nal function is dynamic, and renal with an extensive acute myocardial infarc- some patients would become newly eligi- dysfunction in diabetes is typically pro- tion who developed acute renal failure ble and some who previously qualified gressive (50). Even in the absence of an and died the same day. Given the clinical would now become ineligible). The au- acute event, glomerular function slowly scenario, the authors intimated that aci- thors proposed that if the current practice declines with aging as nephron mass is dosis had occurred because of hemody- is considered safe (and based on the re- lost. The renal thresholds for the accept- namic compromise related to the infarct view by Salpeter et al. [36], this appears to ability of metformin therapy should and not to metformin accumulation. In a be so), then a switch to an eGFR-based cut therefore ideally account for the tempo U.S. study (42) performed in the primary point may be both a more practical and a of CKD progression. The assessment of care practice setting, 4.5% of patients more accurate way to limit metformin ac- renal function in clinical practice occurs treated with metformin had creatinine cess in those with significantly impaired periodically, and the degree of renal dys- levels .1.4 and 1.5 mg/dL in women and renal function. In another British study of function may change appreciably be- men, respectively. Two other studies 12,482 patients with diabetes, an eGFR tween these assessments. Therefore, it is (38,40) of sicker patients admitted to hos- cutoff of 41 mL/min per 1.73 m2 in men important to know how quickly GFR de- pitals in Germany and the U.S. confirmed and 30 mL/min per 1.73 m2 in women clines in the typical spectrum of nephrop- high frequency of metformin use despite resulted in a similar proportion of pa- athy among diabetic patients, particularly various contraindications (73 and 27%, tients having metformin withheld com- when considering metformin therapy. respectively). pared with the serum creatinine threshold Few studies have, however, system- When one considers the imperfect of 1.7 mg/dL (45). The investigators there- atically evaluated the rate of progression reflection of actual renal function by fore proposed the pragmatic eGFR limit of renal dysfunction in the general di- serum creatinine, metformin is likely of 30 mL/min per 1.73 m2 to denote abso- abetic patient population by directly mea- used even more frequently in patients lute contraindication to therapy. suring GFR over time. Some data suggest with impaired GFR than that suggested Limited data specifically address met- that eGFR tends to underestimate the rate by the above studies. In the aforemen- formin’s long-term safety in patients with of GFR decline by as much as 28% when tioned U.S. primary practice setting, mild-to-moderate renal failure (46–48). compared with direct measurement (51). where 4.5% of patients were given met- These studies found no increased risks Nevertheless, most of the available data formin despite creatinine-based contra- in various degrees of renal insufficiency are based on estimations. A recent British indications, 17.7% of women and 13.4% but were limited by small size and signif- population-based cohort study (52) of of men receiving metformin actually had icant methodological shortcomings. Re- 3,431 diabetic patients examined renal an abnormally low eGFR (#60 mL/min cently, an analysis of the Reduction of decline as measured by changes in eGFR per 1.73 m2) (42). Likewise, in another Atherothrombosis for Continued Health (Modification of Diet in Renal Disease). single U.S. center cross-sectional study Registry suggests that the proposed car- The analysis of data collected over 7 years (43), 15.3% of patients with type 2 diabe- diovascular benefits of metformin may demonstrated that the rate of progres- tes and eGFR ,60 mL/min per 1.73 m2 extend to patients with established ath- sion was lowest among individuals who were receiving metformin. Such frequent erosclerosis and moderate CKD (49). were normoalbuminuric (0.3% or ~0.2 “inappropriate” use of metformin in pa- In this large, observational study of mL/min per 1.73 m2 decline in eGFR per tients is further suggested by data from .19,000 subjects with a history of ath- year), intermediate in those with micro- the National Health and Nutrition Exam- erothrombotic disease, 1,572 patients albuminuria (1.5% or ~1.2 mL/min per ination Survey (1999–2006) (43). Among were using metformin with eGFR 30–60 1.73 m2 per year), and highest in those individuals with eGFR ,60 mL/min per mL/min per 1.73 m2. After adjustment for with macroalbuminuria (5.7% or 4.5 1.73 m2 and diabetes, 32.2% were treated baseline factors and propensity score, mL/min per 1.73 m2 per year). In a large with metformin and had a normal creati- metformin use was associated with a sig- Dutch study (53), eGFR fell by 0.5 mL/min nine level (,1.5 mg/dL), whereas 13.4% nificant reduction in 2-year mortality in per 1.73 m2 per year in the general pop- were treated with metformin despite a the overall population (hazard ratio 0.76 ulation but to a greater extent in those with frankly elevated creatinine level (.1.5 [95% CI 0.65–0.89]), including in those hypertension,diabetes,andmacroalbumin- mg/dL). The use of metformin in mild- with moderate CKD (0.64 [0.48–0.86]). uria (1.9 mL/min per 1.73 m2 per year). to-moderate CKD clearly is not at all un- However, lack of information with re- Based on these and other results (31), av- common. spect to the duration of metformin use erage annual progression of renal dysfunc- Two studies have attempted to trans- and HbA1c,aswellastheobservational tion in diabetes appears to be in the late creatinine into corresponding eGFR nature of the study, require further range of 21to24 mL/min per 1.73 m2

1434 DIABETES CARE, VOLUME 34, JUNE 2011 care.diabetesjournals.org Lipska, Bailey, and Inzucchi of filtration capacity, dependent in part Table 1—Proposed recommendations for use of metformin based on eGFR on other risk factors and the use of reno- protective therapies. The decline is slow, eGFR level (mL/min per 1.73 m2) Action but, importantly, the majority of patients in these studies had normal kidney func- $60 No renal contraindication to metformin tion at the outset. There is less information Monitor renal function annually available on diabetic patients with CKD. It ,60 and $45 Continue use should also be noted that diabetes places Increase monitoring of renal function (every 3–6months) an individual at increased risk for other ,45 and $30 Prescribe metformin with caution causes of renal disease (54). Thus, all di- Use lower dose (e.g., 50%, or half-maximal dose) abetic patients, especially those with CKD, Closely monitor renal function (every 3 months) may be at risk for more rapid decline in Do not start new patients on metformin their renal function or acute kidney injury. ,30 Stop metformin Despite these appropriate concerns, Additional caution is required in patients at risk for acute kidney injury or with anticipated significant most of the available data would suggest fluctuations in renal status, based on previous history, other comorbidities, or potentially interacting med- that, on average, eGFR declines slowly in ications. diabetes, although it can be accelerated to some degree in the presence of albuminuria. If eGFR is calculated annually (and be updated. These recommendations CKD, these recommendations would more frequently in those at high risk for should include eGFR thresholds that are have a profound impact on clinical prac- deterioration in renal function), it is un- generally consistent with the National tice. Such proposals, being consistent likely that patients will experience Institute for Health and Clinical Excel- with the sentiments of several well- changes in their eGFR levels large enough lence guidelines in the U.K. and those respected national guideline committees, to rapidly alter the safety of metformin endorsed by the Canadian Diabetes As- should be reviewed by professional med- therapy. sociation and the Australian Diabetes ical organizations in the U.S., such as the Society (Table 1). Metformin may be American Diabetes Association. If a con- CONCLUSIONS AND continued (or initiated) with eGFR ,60 sensus emerges, perhaps the Food and RECOMMENDATIONS—Although mL/min per 1.73 m2, but renal function Drug Administration might reconsider metformin is eliminated renally, and ac- should be monitored closely (every 3–6 the current metformin-prescribing guide- cumulation may conceivably lead to lactic months). The dose of metformin should lines, which, like those of other non- acidosis, there currently is limited sys- be reviewed and reduced (e.g., by 50% or branded compounds, tend to remain tematic evidence to substantiate contin- to half-maximal dose) in those with eGFR static despite emerging evidence and ued reliance on the creatinine-based ,45 mL/min per 1.73 m2, and renal func- changes in clinical care. contraindications in use in the U.S. In- tion should be monitored closely (every In the future, more research in this deed, in the modern era of eGFR, this 3 months). Metformin should not be ini- important area is needed, including pro- measure of glomerular filtration appears tiated in patients at this stage, however. spective, randomized trials of metformin to give a more reliable estimate of renal The drug should be stopped once eGFR at varying degrees of renal impairment dysfunction. Metformin-associated lactic falls to ,30 mL/min per 1.73 m2.Addi- and/or closer examination of registries of acidosis is exceedingly rare based on the tional caution is required in patients with CKD patients receiving metformin. available literature, and even though the anticipated significant fluctuations in re- use of metformin has not been compre- nal status or those at risk for abrupt de- hensively assessed in individuals with terioration in kidney function, based on Acknowledgments—No potential conflicts of CKD, there is extensive evidence that previous history, other comorbidities, al- interest relevant to this article were reported. this agent often is used without adverse buminuria, and medication regimen (e.g., K.J.L. researched data, wrote the manu- effects in those with mildly to moderately potent diuretics or nephrotoxic agents). script, reviewed and edited the manuscript, reduced renal function. In the context of Without question, such a treatment and contributed to discussion. C.J.B. and S.E.I. reviewed and edited the manuscript and rising concerns regarding other glucose- program could not be implemented with- contributed to discussion. lowering therapies (55), safety restric- out meticulous clinical follow-up, clear tions over the use of metformin in this communication with patients regarding population may result in the drug being risks and benefits of therapy, and adher- References stopped prematurely and unnecessarily in ence to frequent monitoring. The plan, 1. Herrington WG, Levy JB. Metformin: ef- some patients. This may lead to significant therefore, should be modified in patients fective and safe in renal disease? Int Urol deterioration in glycemic control and/or with suboptimal adherence to medical Nephrol 2008;40:411–417 the need for other glucose-lowering medi- instructions or regular follow-up. It is 2. McCormack J, Johns K, Tildesley H. cations, which present their own safety Metformin’s contraindications should be clear that vigilance would be required so – concerns. An evidence-based approach that cases of lactic acidosis do not emerge contraindicated. CMAJ 2005;173:502 504 to prescribing metformin in this group ap- because of inappropriate use of metfor- 3. Lalau JD. Lactic acidosis induced by metformin: incidence, management and preven- pears warranted, taking into account the min in patients with more advanced and/ tion. Drug Saf 2010;33:727–740 current pervasive use of eGFR in clinical or unstable CKD. 4. Bailey CJ. Metformin: effects on micro and care. Given the frequency with which clini- macrovascular complications in type 2 We therefore suggest that the current cians must decide whether to continue or diabetes. Cardiovasc Drugs Ther 2008;22: guidelines for metformin use in the U.S. initiate metformin in mild-to-moderate 215–224 care.diabetesjournals.org DIABETES CARE, VOLUME 34, JUNE 2011 1435 Metformin in renal insufficiency

5. Gerstein HC, Miller ME, Byington RP, 17. Schäfer G. Biguanides: a review of history, of glomerular filtration rate estimators in et al.; Action to Control Cardiovascular pharmacodynamics and therapy. Diabetes subjects with type 2 diabetes. Diabetes Care Risk in Diabetes Study Group. Effects of Metab 1983;9:148–163 2008;31:47–49 intensive glucose lowering in type 2 diabetes. 18. Bailey CJ, Turner RC. Metformin. N Engl 33. Lalau JD, Race JM. Lactic acidosis in N Engl J Med 2008;358:2545–2559 J Med 1996;334:574–579 metformin-treated patients: prognostic 6. Patel A, MacMahon S, Chalmers J, et al.; 19. Marchetti P, Benzi L, Cecchetti P, et al. value of arterial lactate levels and plasma ADVANCE Collaborative Group. Inten- Plasma biguanide levels are correlated metformin concentrations. Drug Saf 1999; sive blood glucose control and vascular with metabolic effects in diabetic patients. 20:377–384 outcomes in patients with type 2 diabetes. Clin Pharmacol Ther 1987;41:450–454 34. Stades AM, Heikens JT, Erkelens DW, N Engl J Med 2008;358:2560–2572 20. Oates NS, Shah RR, Idle JR, Smith RL. Holleman F, Hoekstra JB. Metformin and 7. Duckworth W, Abraira C, Moritz T, et al.; Influence of oxidation polymorphism on lactic acidosis: cause or coincidence? A VADTInvestigators.Glucosecontrolandvas- phenformin kinetics and dynamics. Clin review of case reports. J Intern Med 2004; cular complications in veterans with type 2 Pharmacol Ther 1983;34:827–834 255:179–187 diabetes. N Engl J Med 2009;360:129–139 21. Bailey CJ, Nattrass M. Treatment: met- 35. Lalau JD, Lacroix C, Compagnon P, et al. Role 8. UK Prospective Diabetes Study (UKPDS) formin. Baillieres Clin Endocrinol Metab of metformin accumulation in metformin- Group. Effect of intensive blood-glucose 1988;2:455–476 associated lactic acidosis. Diabetes Care control with metformin on complications 22. KDOQI. Clinical practice guidelines and 1995;18:779–784 in overweight patients with type 2 diabetes clinical practice recommendations for 36. Salpeter SR, Greyber E, Pasternak GA, (UKPDS 34). Lancet 1998;352:854–865 diabetes and chronic kidney disease. [In- Salpeter Posthumous EE. Risk of fatal and 9. Holman RR, Paul SK, Bethel MA, Matthews ternet]. Available from http://www.kidney. nonfatal lactic acidosis with metformin DR, Neil HA. 10-year follow-up of intensive org/professionals/KDOQI/guideline_ use in type 2 diabetes mellitus. Cochrane glucose control in type 2 diabetes. N Engl diabetes/guide2.htm. Accessed 5 December Database Syst Rev 2010;(4):CD002967 J Med 2008;359:1577–1589 2010 37. Misbin RI. The phantom of lactic acidosis 10. DeFronzo RA, Goodman AM; The Multi- 23. Canadian Diabetes Association. Clinical due to metformin in patients with diabetes. center Metformin Study Group. Efficacy of practice guidelines [Internet]. 2008. Available Diabetes Care 2004;27:1791–1793 metformin in patients with non-insulin- from http://www.diabetes.ca/files/cpg2008/ 38. Holstein A, Nahrwold D, Hinze S, Egberts dependent diabetes mellitus. N Engl J Med cpg-2008.pdf. Accessed 5 December 2010 EH. Contra-indications to metformin ther- 1995;333:541–549 24. National evidence based guidelines for apy are largely disregarded. Diabet Med 11. Bodmer M, Meier C, Krähenbühl S, Jick blood glucose control in type 2 diabetes. 1999;16:692–696 SS, Meier CR. Metformin, sulfonylureas, [Internet]. Available from http://www.nhmrc. 39. Horlen C, Malone R, Bryant B, et al. Fre- or other antidiabetes drugs and the risk of gov.au/_files_nhmrc/file/publications/ quency of inappropriate metformin pre- lactic acidosis or hypoglycemia: a nested synopses/di19-diabetes-blood-glucose- scriptions. JAMA 2002;287:2504–2505 case-control analysis. Diabetes Care 2008; control.pdf. Accessed 5 December 2010 40. Calabrese AT, Coley KC, DaPos SV, 31:2086–2091 25. Wilcock C, Bailey CJ. Accumulation of Swanson D, Rao RH. Evaluation of pre- 12. Nathan DM, Buse JB, Davidson MB, et al.; metformin by tissues of the normal and scribing practices: risk of lactic acidosis American Diabetes Association; European diabetic mouse. Xenobiotica 1994;24: with metformin therapy. Arch Intern Med Association for Study of Diabetes. Medical 49–57 2002;162:434–437 management of hyperglycemia in type 2 26. Scheen AJ. Clinical pharmacokinetics of 41. Emslie-Smith AM, Boyle DI, Evans JM, diabetes: a consensus algorithm for the metformin. Clin Pharmacokinet 1996;30: Sullivan F, Morris AD; DARTS/MEMO initiation and adjustment of therapy: a 359–371 Collaboration. Contraindications to met- consensus statement of the American 27. Tucker GT, Casey C, Phillips PJ, Connor formin therapy in patients with type 2 Diabetes Association and the European H, Ward JD, Woods HF. Metformin ki- diabetes: a population-based study of Association for the Study of Diabetes. netics in healthy subjects and in patients adherence to prescribing guidelines. Dia- Diabetes Care 2009;32:193–203 with diabetes mellitus. Br J Clin Pharma- bet Med 2001;18:483–488 13. Rodbard HW, Jellinger PS, Davidson JA, col 1981;12:235–246 42. Kennedy L, Herman WH; GOAL A1C et al. Statement by an American Associa- 28. Sambol NC, Chiang J, Lin ET, et al. Kid- Study Team. Renal status among patients tion of Clinical Endocrinologists/American ney function and age are both predictors using metformin in a primary care setting. College of Endocrinology consensus panel of pharmacokinetics of metformin. J Clin Diabetes Care 2005;28:922–924 on type 2 diabetes mellitus: an algorithm for Pharmacol 1995;35:1094–1102 43. Vasisht KP, Chen SC, Peng Y, Bakris GL. glycemic control. Endocr Pract 2009;15: 29. Lalau JD, Vermersch A, Hary L, Andrejak Limitations of metformin use in patients 540–559 M, Isnard F, Quichaud J. Type 2 diabetes with kidney disease: are they warranted? 14. National Institute for Health and Clinical in the elderly: an assessment of metformin Diabetes Obes Metab 2010;12:1079– Excellence. The Management of Type 2 (metformin in the elderly). Int J Clin 1083 Diabetes: 2010 NICE Guidelines [Internet]. Pharmacol Ther Toxicol 1990;28:329– 44. Warren RE, Strachan MW, Wild S, London, U.K., National Institute for Health 332 McKnight JA. Introducing estimated glo- and Clinical Excellence, 2010. Available 30. Frid A, Sterner GN, Löndahl M, et al. merular filtration rate (eGFR) into clinical from http://www.nice.org.uk/nicemedia/ Novel assay of metformin levels in pa- practice in the UK: implications for the live/12165/44320/44320.pdf. Accessed tients with type 2 diabetes and varying use of metformin. Diabet Med 2007;24: 21 October 2010 levels of renal function: clinical recom- 494–497 15. Knowler WC, Barrett-Connor E, Fowler mendations. Diabetes Care 2010;33:1291– 45. Shaw JS, Wilmot RL, Kilpatrick ES. Estab- SE, et al.; Diabetes Prevention Program 1293 lishing pragmatic estimated GFR thresholds Research Group. Reduction in the in- 31. Rossing P, Rossing K, Gaede P, Pedersen to guide metformin prescribing. Diabet Med cidence of type 2 diabetes with lifestyle O, Parving HH. Monitoring kidney func- 2007;24:1160–1163 intervention or metformin. N Engl J Med tion in type 2 diabetic patients with in- 46. Rachmani R, Slavachevski I, Levi Z, Zadok 2002;346:393–403 cipient and overt diabetic nephropathy. B, Kedar Y, Ravid M. Metformin in patients 16. American Diabetes Association. Standards Diabetes Care 2006;29:1024–1030 with type 2 diabetes mellitus: reconsideration of medical care in diabetes: 2010. Diabetes 32. Chudleigh RA, Dunseath G, Peter R, et al. of traditional contraindications. Eur J Intern Care 2010;33(Suppl. 1):S11–S61 Influence of body weight on the performance Med 2002;13:428

1436 DIABETES CARE, VOLUME 34, JUNE 2011 care.diabetesjournals.org Lipska, Bailey, and Inzucchi

47. Lim VC, Sum CF, Chan ES, Yeoh LY, Lee with diabetes and atherothrombosis. Arch 53. van der Velde M, Halbesma N, de Charro FT, YM, Lim SC. Lactate levels in Asian pa- Intern Med 2010;170:1892–1899 et al. Screening for albuminuria identi- tients with type 2 diabetes mellitus on 50. Holt RIG, Goldstein BJ. Textbook of Di- fies individuals at increased renal risk. metformin and its association with dose of abetes. Oxford, U.K., Wiley-Blackwell, J Am Soc Nephrol 2009;20:852–862 metformin and renal function. Int J Clin 2010 54. Pham TT, Sim JJ, Kujubu DA, Liu IL, Pract 2007;61:1829–1833 51. Xie D, Joffe MM, Brunelli SM, et al. A Kumar VA. Prevalence of nondiabetic renal 48. Connolly V, Kesson CM. Metformin comparison of change in measured and disease in diabetic patients. Am J Nephrol treatment in NIDDM patients with mild estimated glomerular filtration rate in pa- 2007;27:322–328 renal impairment. Postgrad Med J 1996; tients with nondiabetic kidney disease. Clin 55. U.S. Food and Drug Administration. 72:352–354 J Am Soc Nephrol 2008;3:1332–1338 Guidance for Industry Diabetes Mellitus: 49. Roussel R, Travert F, Pasquet B, et al.; Re- 52. Hoefield RA, Kalra PA, Baker PG, et al. Evaluating Cardiovascular Risk in New An- duction of Atherothrombosis for Contin- The use of eGFR and ACR to predict decline tidiabetic Therapies to Treat Type 2 Di- uedHealth(REACH)RegistryInvestigators. in renal function in people with diabetes. abetes. Rockville, MD, U.S. Food and Metforminuseandmortalityamongpatients Nephrol Dial Transplant 2011;26:887–892 Drug Administration, 2008

care.diabetesjournals.org DIABETES CARE, VOLUME 34, JUNE 2011 1437