Keryx Biopharmaceuticals' (NASDAQ:KERX) Sulonex (sulodexide) is an orally available heparinoid (80% heparan sulfate, 20% dermatan sulfate) currently in phase III/IV trials for use in diabetic nephropathy. Though its mechanism of action in diabetic nephropathy is uncertain, it is hypothesized that sulodexide – a gylcosaminoglycan [GAG] – restores the anionic charge in the mesangium and glomerular basement membrane [GBM] of the kidney’s nephron. As GAG loss – and the subsequent decrease in anionic charge of the GBM – is one of the hallmarks of diabetic nephropathy, it is felt that sulodexide may, in effect, act as a GAG replacement in the GBM, thereby restoring the integrity of the GBM damaged by poor gylcemic control.
Other hypotheses for sulodexide’s mechanism of action include inhibition of TGF-beta(1), a cytokine known to cause fibrosis in the GBM and nephron, and inhibition of heparanase enzymes, which may degrade GAG products in the kidney’s glomeruli. A pivotal phase III trial has begun, evaluating the ability of sulodexide to promote regression of the urinary albumin/creatinine ratio (UACR) in patients with type 2 diabetes and microalbuminuria. In addition, there is an on-going phase IV trial evaluating sulodexide’s effect on time to doubling of serum creatinine, or time to development of end-stage renal disease, in patients with type 2 diabetes and overt proteinuria.
Elements of the Collaborative Study Group Phase II Trial that Argue Against the Generation of Positive Phase III Data
- The baseline characteristics of the sulodexide 200 mg/day group and the placebo group were not equivalent, and were distributed in a manner that favored regression of proteinuria in the sulodexide group. The sulodexide 200 mg/day group had significantly lower total cholesterol than both the placebo and the 400 mg/day groups. This is an important point, given that lower cholesterol levels have been demonstrated to have a salutary effect on the spontaneous regression of microalbuminuria, even in patients who do not have overt dyslipidemia. In this Collaborative Study Group trial, the median cholesterol level for the 200 mg/day treatment group was 160.2; the placebo and 400 mg/day group had median levels of 183.8 and 177.8, respectively. Given that only the 200 mg/day group demonstrated a regression in proteinuria during the study period – and the 400 mg/day group did not, despite previous belief that there is a linear dose-response relationship between sulodexide and proteinuria – it is plausible that the clinical response observed in the 200 mg/day group is more reflective of the baseline lipid profile of the cohort, and may have had nothing to do with the treatment medication. The plausibility of this hypothesis gains credence when it is considered that cholesterol profile was the only clinically significant baseline characteristic that differed between the study groups, and this happened to be the only group to demonstrate a clinical response (though, it should also be noted, even this response was not clinically significant). Two other points should be considered in this regard: 1) Not only have lower total cholesterol levels been demonstrated to effect a higher rate of spontaneous regression of albuminuria, but increased levels of total cholesterol have likewise been demonstrated (and accepted by the Work Group of the Kidney Disease Outcomes Quality Initiative [KDOQI]) to increase albuminuria and accelerate the progression of diabetic nephropathy; in this manner, not only was the 200 mg/day group provided a theoretic advantage, but the other groups were placed at an additional theoretic disadvantage. And, 2) triglyceride levels, like cholesterol levels, have similarly been shown to promote spontaneous regression of albuminuria in diabetics when they are low, and promote albuminuria when high – interestingly, the sulodexide 200 mg/day group also had lower triglyceride levels than the other groups, though this difference was reported as statistically insignificant.
- The sulodexide 200 mg/day cohort also had less than half as many black patients as the placebo group (10% vs. 21.3%), and more Caucasians (90% vs. 78.7%). Though the investigators report that this difference in ethnic profile was not statistically significant, I believe this difference merits consideration: One of the features of this trial was a run-in period in which patients were titrated to a maximum recommended dose of either an ACE inhibitor or angiotensin receptor blocker. Both of these drug classes have been demonstrated to decrease proteinuria in patients with diabetic nephropathy, but these drugs do not have equal efficacy in all ethnic groups – most notably, they are less effective in African Americans, for diabetic nephropathy as well as for hypertension and heart failure. It is therefore conceivable that some of the differences observed in the 200 mg/day group can be attributed to a cohort profile that would be expected to be more responsive to other background therapies mandated by the trial’s design. Another point to consider is the greater susceptibility of minorities to diabetic kidney disease: though different ethnic groups may have similar rates of microalbuminuria, the rate of progression is higher in blacks and other non-Caucasians. To be fair, it should be noted that the 400 mg/day group had a similar ethnic profile as the 200 mg/day group, but did not fare as well – although neither its cholesterol nor triglyceride profiles (both important factors for spontaneous regression of proteinuria) were as favorable as those in the 200 mg/day group.
- There were four times as many serious adverse events in the sulodexide 200 mg/day group (16) as compared to placebo (4). In addition, there were two deaths recorded during the study period, and both patients were in a sulodexide group. This non-reassuring data was explained away by the investigators’ inference that neither the fatal nor the non-fatal SAEs were probably related to the study drug, and their contention that this trial was not designed to evaluate drug safety. However, it should be noted that the greatest risk of morbidity and mortality to a patient with diabetic nephropathy is cardiovascular disease – these patients are more likely to succumb to cardiovascular disease than they are to their kidney disease – and 50% of the SAEs in this trial were cardiac events. So whereas the researchers may be correct in their assertion that sulodexide did nothing to directly trigger a cardiac event, it does seem possible that sulodexide was inferior to placebo in preventing a clinical endpoint that is much more important than albuminuria – namely, cardiovascular morbidity. Though it might be acceptable to dismiss these discrepancies in SAEs in a relatively small phase II study, it will be harder to dismiss such poor outcomes in a phase III trial – where the larger patient recruitment might lend more statistical power to such discrepancies, and trial design will be more sensitive to events. The phase IV trial will similarly be larger and more sensitive to the development of clinical events. It should also be noted that 11% of the patients (10/91) on the study medication had to temporarily discontinue its use due to an adverse event. Only two patients in the entire study were lost to follow-up due to an adverse event – both were in the sulodexide 200 mg/day group.
- The study evaluated urinary albumin/creatinine ratios in a first morning void. Though the Work Group of the KDOQI accepts a first morning UACR as a screening test for microalbuminuria, it does not satisfy the criterion of providing proven clinical benefits, as the impact of microalbuminuria has not been demonstrated on such hard clinical endpoints as end stage renal disease or cardiovascular morbidity/mortality. In the opinion of the Work Group, there is insufficient evidence to assume that lowering albuminuria levels will necessarily lead to improvements in such clinical outcomes as progression to renal failure, cardiovascular events, or death. In addition, more recent information has revealed that a substantial proportion of patients with type 2 diabetes and microalbuminuria (perhaps as high as 50%) spontaneously regress to normoalbuminuria, calling into question the inevitability of kidney disease progression. As a result of all of these points, the Work Group of the KDOQI mandates that potential treatments for diabetic nephropathy demonstrate not only albuminuria reduction, but also reduction in end stage renal disease, cardiovascular events and death, before being considered efficacious. This is relevant to the on-going phase III trial for sulodexide, in the sense that its only primary endpoint is albuminuria reduction – though it conceivably could achieve its endpoint goals, acceptance by the national academy will require much more than a demonstration of albuminuria regression, and so a positive phase III trial alone may not convince the academy. For sure, the concurrent phase IV trial – which is measuring the progression to kidney failure as a primary endpoint – will also have to demonstrate positive data. And if the phase III trial – like the phase II trial before it – demonstrates a decrease in albuminuria while simultaneously demonstrating an increase in cardiac events, it is safe to assume that sulodexide will be rejected as a therapy for diabetic nephropathy.
- The clinical benefit that the investigators report in the sulodexide 200 mg/day group did not achieve statistical significance. This is the most obvious shortcoming of this phase II trial – though the decrease in UACR in the 200 mg/day sulodexide group came close to achieving statistical significance, it fell just short (p=.075).
- The phase III trial will have a longer run-in period on ACE inhibitors or angiotensin receptor blockers than the phase II trial. The run-in period on ACEIs or ARBs was 60 days during the phase II trial; for phase III, this run-in period will be 120 days. This will make the prospect of positive, clinically significant data more difficult in two ways: 1) with an extra two months of ACEI or ARB therapy, it might be expected that the placebo group will, in essence, be given greater opportunity to lower their UACR during the study period, and; 2) all study participants might be expected to have lower baseline UACR at the end of the run-in, making a therapeutic success (defined as a >50% reduction in UACR, or a normalization of albuminuria with a >25% reduction in albuminuria) in the treatment groups a much more difficult clinical endpoint to achieve.
- Though previous pilot studies, including a European soil phase II trial, had shown linear, dose-dependent responses to sulodexide in the treatment of diabetic nephropathy, in this study the sulodexide 400 mg/day group fared worse than the 200 mg/day group. The investigators explain this unusual, parabolic curve with three possible hypotheses. One, they suggest that the therapeutic effect of sulodexide may plateau at 200 mg/day; however, this study did not demonstrate a plateau effect, but a decrease in clinical response, comparable to placebo, at higher doses. Two, the investigators propose a bio-feedback mechanism regarding anionic charges on the GBM at higher sulodexide levels; certainly it is possible that sulodexide, whatever its mechanism of action, may be a mixed agonist-antagonist of some glomerular enzyme. Third, they propose that the absence of a stronger trend at 400 mg/day was due to chance variation; if this is true, however, then certainly the same could be said for the clinical benefit observed in the 200 mg/day group, especially since that cohort was the only outlier of the three groups in the study.
Elements of the Di.N.A.S. Phase II Trial that Do Not Support the Generation of Positive Phase III Data
- The investigators chose urinary albumin excretion rate as an endpoint, without correcting for urinary flow rates and/or kidney function by measuring the ratio of urinary albumin to urinary creatinine. It is possible that urinary albumin excretion would go down simply as a function of decreased urine production, and a failing kidney might actually begin to lose less protein simply as a function of its decreased capacity to produce urine. In this manner, it is conceivable that decreased albumin excretion was not the sign of a kidney with improving function, but a sign of a kidney that was losing its capacity to maintain appropriate urine volumes. This design flaw could have been easily corrected by concurrent measures of urinary creatinine, and then generating an albumin/creatinine ratio. In fact, a UACR (as opposed to a simple albumin excretion rate) is the preferred method of evaluation for albuminuria as recommended by the Work Group of the KDOQI, for the additional reason that timed collections are prone to inaccuracy. Furthermore, urinary albumin excretion demonstrates significant (approximately 40%) intraindividual variation.
- The study did not evaluate 24-hour urine specimens, but overnight, 8-hour timed specimens. Overnight urine specimens are considered sufficient for evaluation of proteinuria if the protein is measured as a ratio to creatinine – when this is done, a first morning UACR has an approximately 80% correlation with 24-hour urine collections. However, overnight 8-hour specimens are considered extremely inaccurate without this correction. Besides the significant inaccuracies associated with overnight 8-hour specimen collection, overnight specimens reflect both decreased urinary flow and decreased protein excretion – a factor whose impact would have been reduced had an albumin/creatinine ratio been measured. Unlike the Collaborative Study Group trial, however, this change in diurnal urinary excretion rates was not corrected by generating an albumin/creatinine ratio, which makes the measurements in the Di.N.A.S. trial less reliable.
- Not only was an insufficient biomarker – albumin excretion rate – measured in the Di.N.A.S. study as a primary endpoint, it is a completely distinct endpoint from those being measured in the phase III and IV trials. Though the data generated by the Di.N.A.S. trial was positive, it should be noted that not only was an inferior endpoint measured, but that endpoint is not even common to the pivotal trials of sulodexide. The pivotal phase III trial will be measuring urinary albumin/creatinine ratio, which has poor correlation with albumin excretion rates derived from overnight samples. And the phase IV trial will evaluate time to doubling of serum creatinine, or time to end stage renal disease – both of which have even poorer correlation with albumin excretion rates. Practically speaking, the Di.N.A.S. study is of no predictive value for evaluating the prospects of the pivotal sulodexide trials.
Limitations of Previous Sulodexide Pilot Studies That Have Reported Positive Data
Though the phase II Collaborative Study Group trial was hampered by differences in baseline characteristics of the study groups; the inability to achieve statistical significance in the measured endpoint; and the surprise observation that higher doses of sulodexide resulted in endpoints that trended toward placebo, previous pilot studies of sulodexide did generate statistically significant, positive data. It is important to address these positive-result pilot studies, and identify any shortcomings in their design. The following are three examples of positive pilot studies, and examples of their design flaws that would lead to dismissal of their data:
Dedov I et al.: A randomized, controlled study of sulodexide therapy for the treatment of diabetic nephropathy. Nephrol Dial Transpl 1997.
This study demonstrated a decrease in albumin excretion rates in sulodexide-treated patients. However:
- The control groups, particularly those with macroalbuminuria, had much higher systolic blood pressure at baseline than the treatment groups – lower systolic blood pressure is a known salutary factor for spontaneous regression of proteinuria, particularly in type 1 diabetics (everyone in this study had type 1 diabetes).
- The typical control patient was in his 30s, while the typical treatment patient was in his 20s – lower age is also a known salutary factor for spontaneous regression of proteinuria, particularly in type 1 diabetics.
- Albumin excretion rate, and not UACR, was the measured endpoint.
- The trial had no standardization of blood pressure medications.
- The trial was three weeks long, with a six week follow-up.
Poplawska A et al.: Effect of gylcosaminoglycans on urinary albumin excretion in insulin-dependent diabetic patients with micro- or macroalbuminuria. Diabetes Res Clin Pract 1997.
This study demonstrated a decrease in albumin excretion rates in sulodexide-treated patients. However:
- There was no placebo group in this trial
- Inclusion criteria for the type 1 diabetics in this trial was a diagnosis of at least three years’ duration – this short time frame would not be accepted by the Work Group of the KDOQI as a long enough period in which a type 1 diabetic would acquire nephropathy. Furthermore, spontaneous regression of proteinuria is positively correlated to short length of disease.
- All of the patients in this study, by inclusion criteria, had a HgbA1c of <8% -- this is a magic number, for HgbA1c<8% has been demonstrated to be an important factor in the spontaneous regression of proteinuria, particularly in type 1 diabetics.
- No one in this study was allowed to be on an ACE inhibitor.
- Albumin excretion rate, and not UACR, was the measured endpoint.
- The trial was only 31 days long.
Sulikowska B et al.: Effect of sulodexide on albuminuria, NAG excretion and glomerular filtration response to dopamine in diabetic patients. Am J Nephrol 2006
This study demonstrated decreased albuminuria from 24-hour collections, as well as decreased excretion of a biomarker (NAG) that is associated with progressive diabetic nephropathy. However:
- Every subject in this study had type 1 diabetes, and was very young (roughly half the age of the typical patient in the phase II sulodexide study).
- The subjects in this study had a very low penetration of ACE inhibitors in their background therapy.
- Though a 24-hour urinary albumin was measured, and this has improved accuracy when compared to overnight albumin excretion rates, the specimens were not corrected by a creatinine ratio.
- The study was randomized, but there was no placebo in the control group, and no blinding.