How the latest evidence from clinical research informs patient care

Abstract

Citation: Raggi P, Chertow GM, Torres PU, et al. The ADVANCE Study: a randomized study to evaluate the effects of cinacalcet plus low-dose vitamin D on vascular calcification in patients on hemodialysis. Nephrol Dial Transplant. 2011;26:1327–1339.

Analysis: Cardiovascular disease remains a significant burden in hemodialysis patients, and coronary calcification has been shown to be an independent risk factor for both cardiovascular morbidity and all-cause mortality in this group.1, 2 Furthermore, coronary calcification tends to progress rapidly once established.3 Several studies have looked at the effects of phosphate binders, calcium intake, and statins on coronary calcification progression and mortality, with some evidence that lower calcium intake along with lower phosphorus levels slows calcification and improves outcomes.2, 4, 5 This study by Raggi and colleagues looks at cinacalcet and its effect on coronary and cardiac valve calcifications in patients with secondary hyperparathyroidism.

This prospective controlled trial randomized 360 hemodialysis patients with secondary hyperparathyroidism to therapy with cinacalcet and low dose vitamin D or to conventional therapy with vitamin D sterols alone. Patients were included if they had moderate to severe hyperparathyroidism and a baseline Agatston coronary artery and valvular calcification scores at or above 30. Calcification scores were measured at 28 and 52 weeks, and all patients were treated with calcium-based phosphate binders. The results demonstrated a statistically significant slower progression of scores in the aortic valve, with consistently less of an increase (although not reaching statistical significance) in calcification scores of the coronary arteries, aorta, and mitral valves in the cinacalcet group. Patients in the cinacalcet group had a greater decrease in serum PTH, calcium, and phosphorus as well as lower vitamin D requirements. The authors concluded that in patients with moderate to severe hyperparathyroidism, cinacalcet with vitamin D may attenuate coronary and cardiac valve calcification.

Validity and threats to validity: This large multicenter trial included a diverse patient population broadly representative of the full diversity of age, race, and dialysis vintage in the prevalent dialysis population. Patients were stratified based on calcification score before randomizing, the latter in a 1:1 fashion to the two treatment arms. Stratification is often beneficial in preventing imbalance between groups for factors known to influence treatment responsiveness and overall prognosis (such as baseline calcification scores). It also prevents a type I error (rejecting the null hypothesis when it is true), and it improves power in trials with fewer than 400 patients.6 Additional strengths of the study include the validation of calcification scoring by using a single, blinded reader, with re-evaluations of samples to assess variability. Independent assessments were also used in a subset of scans, showing a high correlation coefficient. All patients were on calcium based binders. As the previously referenced studies have shown the possible contribution of calcium-based binders to coronary calcification when compared to non-calcium based binders, maintaining consistency of the regimen clarifies the findings with regards to the effects of cinacalcet.

This is a very robustly designed, adequately powered and rigorously conducted RCT. Several features of the study, however, deserve some consideration. First, more than 25% of each group lacked a CT scan for calcification scoring, either at the beginning or end of the study, reducing the precision of the results. Additionally, 19 patients in the cinacalcet group and 12 patients in the control group did not complete the per protocol regimen and were not included in the analysis (drop-out rate of < 10%). While the efficacy analysis (as treated) likely provides a compelling picture of the potential magnitude of effect of the intervention, excluding data on a sizeable fraction of the patients (> 20%) may significantly distort the distribution of the baseline prognostic characteristics achieved by random allocation. In this study, despite the small drop-out rate, there were some imbalances in clinical features that might have had prognostic implications. The control group had a slightly higher number of women, and although the stratification was carried out prior to randomization, patients with the highest calcification scores tended to be men. Also, nearly twice as many patients in the control group had a history of congestive heart failure. Whether or not these clinical characteristics could have influenced the outcome and whether or not an ascertainment bias may have been introduced cannot be entirely excluded, but the risk of these is reduced by the masked allocation and low drop-out rate.

There was a definite, consistent, and progressively increasing (with time) separation of the groups over the 52 weeks of follow-up when compared to baseline and 26 weeks. Some of the observed trends that did not quite reach statistical significance would likely have reached statistical significance had the study been carried out for a longer period of time.

Clinical bottom line: This study provides compelling support for the view that cinacalcet, possibly through its effect on calcium load, will quite favorably influence the rate of coronary calcification progression. This study is, however, unable to answer some important additional clinical questions. Is an intervention initiated in prevalent dialysis patients who already demonstrate calcification too late to achieve optimal benefit on hard patient-centered outcomes such as survival and reduced rates of cardiovascular events? What combination of therapies with vitamin D, phosphate binders (calcium versus non-calcium based binders) and cinacalcet would result in optimal outcomes? Whether or not cinacalcet therapy also improves overall morbidity and mortality remain to be seen. This question may be answered by the EVOLVE (Evaluation of Cinacalcet Therapy to Lower Cardiovascular Events) study within the next 2 years.

Citation: Holdaas H, Holme I, Schmieder RE, et al. Rosuvastatin in diabetic hemodialysis patients. J Am Soc Nephrol. 2011; 22: 1335–1341, 2011.

Analysis: A wealth of robust clinical trials evidence in many diverse populations demonstrates that the “statins” including rosuvastatin modify cardiovascular risk and improve survival when administered to patients with dyslipidemia and at higher risk for future cardiovascular events. However, the AURORA Trial (A Study to Evaluate the Use of Rosuvastatin in Subjects on Regular Hemodialysis: An Assessment of Survival and Cardiovascular Events) confirmed earlier evidence that for prevalent hemodialysis patients in general, no cardiovascular risk reduction with statin treatment could be demonstrated during long-term follow-up.7, 8 Recently, two post-hoc subgroup analyses have been reported on data from the two largest randomized controlled clinical trials, the AURORA trial and the Die Deutsche Diabetes Dialysis Study 4 (4-D) trial.8 These analyses support the conclusion that a subset of the hemodialysis patients with high prognostic risk profiles might benefit substantially from statin therapy even if such benefit cannot be demonstrated in the hemodialysis population as a whole.9 We will review the post-hoc analysis of the AURORA trial in this literature watch.

This analysis, with modification of end points, demonstrated a significant cardiac events (cardiac deaths and non-fatal MI) reduction of 32% among diabetic hemodialysis patients taking 10 mg rosuvastatin daily, a non-significant reduction of 16.2% in the composite outcome of stroke, cardiac death and non-fatal MI, and a small increase in the risk of hemorrhagic strokes but no increase in all strokes. Cardiac events were reduced from 10.8 events/100 patient years to 7.8 events per 100 patient years, (HR 0.68; 95% CI 0.51–0.90) for a number needed to treat (NNT) of 33 patients for one year to prevent one cardiac event.10 Treatment effects adjusted for Beta-blockers, angiotensin converting enzyme/angiotensin receptor blockers or sevelamer usages, serum calcium and phosphate levels were all well matched and no significant interactions with mortality or primary composite endpoints were found. Traditional cardiac risk factors such as LDL-cholesterol, blood pressure (BP) and smoking were not associated with significant cardiac outcomes. The only independent risk factors found to be associated with first cardiac events in the diabetic hemodialysis patients were age (HR 1.023; 95% CI 1.053–1.345; P = 0.018) and high sensitivity C-reactive protein (HR 1.190; 95% CI 1.053–1.345; P = 0.005).

Validity and threats to validity: The current study is a robust post-hoc analysis of the diabetic patient strata from the AURORA trial. Typically, post-hoc analyses are considered to be less informative because they are often seen as “fishing expeditions” and they often rely on subsets of the study population where randomization was not utilized to insure a balance of the baseline prognostic clinical features. The current study by Holdaas and colleagues differs from the typical post-hoc analysis in two very important ways. First, this analysis was conducted to test a predetermined hypothesis; and second, the analysis was conducted on a stratum of 731 diabetic patients who balanced in the two arms of the original trial, 384 assigned to treatment and 343 to control. Furthermore, the analysis was performed on an intention-to-treat population and all diabetic patients originally randomized were included in the final analysis. All events classified as cardiac events were adjudicated by a Clinical Endpoint Committee.

Nevertheless, there were a few potential threats to validity that need to be considered. Differential misclassification of any of the cardiac deaths as non-cardiac might distort the findings. However, a common definition of cardiac deaths was employed and the adjudicators of this outcome were masked to the study patients' treatment assignment (and to their lipid profiles) so that a differential misclassification is unlikely. While a blocked randomization strategy was used to balance allocation by center for the overall trial, one cannot be certain of balance for the diabetic patient subset of this trial, but given the size of this trial, the large number of centers and the absence of predominance by any one center, a center-bias is also unlikely.

Clinical bottom line: The current study, as well as the contemporary post-hoc analysis of the 4-D trial, supports the conclusion that statin therapy may substantially benefit those hemodialysis patients with a higher risk for cardiovascular events and in these populations appears to be safe. The 4-D trial post hoc analysis finds similar benefit amongst diabetic patients with the highest pre-trial LDL levels further extending the findings of the post-hoc analysis of the AURORA.

These two studies illustrate two important concepts when evaluating evidence that can be rigorously exploited to inform best clinical practice. The absence of a significant benefit when reported from a well designed robustly conducted clinical trial does not necessarily mean the lack of benefit for all patients. An underpowered study is simply inconclusive and is not, as often reported, a negative study.11 Even in the case of a well powered study such as both the 4-D and the AURORA trials, the ability of a trial to demonstrate a benefit when one truly exists (power) depends importantly on the baseline risk of the individuals included in the trial. Benefits that might be seen for high-risk patients might be obscured by a more marginal benefit (or even a potential for net harm) amongst the very lowest risk patients if both are included in a clinical trial. Differential benefits for different risk strata of the population can be evaluated using the concept of threshold number needed to treat.12, 13 Any conclusions based on a small numbers of events even if statistically significant can be rather uncertain, just a small number of changes in events can significantly alter the conclusion. Thus, the findings regarding hemorrhagic stroke in the post-hoc AURORA analysis are uncertain even if statistically significant.

The clinical bottom-line, however, most importantly rests on the overall finding of this study, namely that prevalent hemodialysis patients with diabetes, a high-risk group of patients (including potentially those with the highest LDL levels) may benefit substantially from inclusion of statins in their regimen. An NNT of 33 patients can be determined and this number compares favorably with NNTs reported for other widely adopted therapies including statin use in the non-CKD high-risk population.14

Citation: Hanaway MJ, Woodle ES, Mulgaonkar S, et al. (for the INTAC Study Group). Alemtuzumab induction in renal transplantation. N Engl J Med. 2011;364:1909–1919.

Analysis: The INTAC trial published in NJEM in May 2011 is a prospective, open label, multi-centered randomized controlled trial designed to examined the safety and efficacy of alemtuzumab (Campath-1H), an anti-CD52 T-cell and B-cell depleting monoclonal antibody, as an induction agent in renal transplantation compared to current customary induction therapy with either rabbit anti-thymocyte globulin (Thymoglobulin), a lymphocyte depleting polyclonal antibody or basiliximab (Simulect), an IL-2 receptor monoclonal antibody which is non-lymphocyte depleting. After stratification into high and low immunologic risk groups, 501 patients were randomly assigned in a blocked randomization schema prior to their actual transplant to receive either experimental therapy with alemtuzumab or conventional induction therapy. High-risk patients were defined as patients undergoing repeat transplantation, a peak or current value of panel-reactive antibodies of 20% or greater or black race. Patients who received kidneys from expanded criteria donors, from donors without a heartbeat, kidneys with ischemic times exceeding 36 hours, positive cytotoxic or flow-cytometric cross-matches and kidneys from HLA-identical live donors were excluded from the study. Patients assigned to alemtuzumab received a single intravenous 30 mg dose at time of transplantation. Low-risk patients in the conventional arm received intravenous Anti-IL-2r 20 mg at the time of transplantation and another 20 mg dose on day 3, day 4, or day 5. High-risk patients in the conventional arm received intravenous rATG dose of 1.5mg/kg given on day 0, day 1 and day 2 and either day 3 or day 4. All patients received tacrolimus, mycophenolate mofetil or mycophenolic acid as a substitute and 1 gram or less of prednisolone or equivalent glucocorticoid over a period of 5 days. Tacrolimus target whole blood trough levels of 7–14 ng per mL were maintained for the first 90 days after transplantation and 4–12 ng per mL thereafter. Mycophenolate was given at a dose of 2 grams per day and mycophenolic acid was given at 1440 mg per day. The antimetabolites could be adjusted by the investigator on the basis of adverse effects. Steroids were stopped at day 5 after transplantation before discharge from the hospital. Each institution used its own protocols for prophylaxis against CMV, PJP, fungi or bacteria and its own protocols for treatment of acute rejection. The primary endpoint was the rate of biopsy-confirmed acute rejection at 6 months and 12 months. Secondary end points included clinical measures of rejection, measures of efficacy, patient and graft survival, graft function, infection, cancer and metabolic measures. Antibody-mediated rejection was not a pre-specified end point and donor specific antibody data was not collected.

Given the study design of a traditional superiority trial, the authors determined that they would need a total sample of 500 patients to provide an 85% power to detect a significant absolute risk difference of at least 9–11% between alemtuzumab and conventional therapy at 6 and 12 months. Six and 12 month analyses were repeated at 36 months. Analyses were performed on data from study patients who received at least one dose of tacrolimus and one dose on induction therapy in a modified intention-to-treat analysis.

After enrollment, 474 patients underwent transplantation and were, therefore, eligible for the study. There were 139 high-risk recipients; 70 received alemtuzumab and 69 received rATG and 335 low-risk recipients; 164 received alemtuzumab and 171 received Anti- IL-2r. There were no significant differences in baseline patient characteristics between the treatment groups. The alemtuzumab group had a significantly (P = 0.04) larger percentage of male donors than the conventional group. The rate of biopsy-confirmed acute rejection was significantly lower among patients receiving alemtuzumab than among those receiving conventional therapy at 6 and 12 months (3% vs. 15%, P < 0.001, 5% vs. 17%, P < 0.001) after transplantation. The rate of biopsy confirmed acute rejection when re-analyzed at 36 months remained lower with alemtuzumab than with conventional therapy (13% vs. 20%, P = 0.03). In high-risk patients, no significant difference was seen between the treatment groups at 6, 12, and 36 months. There was a significantly higher rate of late biopsy confirmed acute rejection, defined as rejection occurring between 12 and 36 months after transplantation in patients who did not have biopsy confirmed acute rejection within the first 12 months, in the alemtuzumab group as compared to the conventional induction regimen arm (8% vs. 3%, P = 0.03). This finding was not seen in either the low-risk patient group or in the entire high-risk patient group. There was no difference in the patient survival and graft survival or graft function at 3 years. The incidence of serious adverse events related to cancer was higher in the alemtuzumab group (P = 0.03) as compared to the conventional induction group, but the overall rate of adverse events related to cancer was similar in the two treatment groups and given the small number of cancers this result while significant is rather fragile. Among low-risk patients, the rate of serious infectious adverse events was significantly lower with Anti- IL-2r (22% vs. 35%, P = 0.02), as compared to the alemtuzumab group, but the rates of all infectious adverse events did not differ significantly between the two groups. Among high-risk patients, the rate of all infectious adverse events was significantly higher with rATG as compared with alemtuzumab (81% vs. 60%, P = 0.009), but the rates of serious infectious adverse events were similar.

Validity and threats to validity: This study is of clinical importance because it is the first multicenter randomized trial comparing alemtuzumab to conventional induction therapy in renal transplantation, enrolling a large number of patients and, therefore, likely sufficiently powered to demonstrate a clinically meaningful difference between treatment arms if one exists. The conventional induction regimens were Anti-IL-2r in low-risk patients and rATG in high-risk patients. The trial used a narrow definition of “high risk,” excluding some common types of high-risk patients so that this study is not directly applicable to patients similar to those excluded from the trial. While 501 patients were randomized, only 474 patients were included in the primary analysis and, therefore, the study might not have been powered sufficiently to detect a smaller than predicted but clinically meaningful difference, especially if one considers the larger number of subjects required for power necessitated when one considers the two different control treatment regimens and two risk strata separately.

More importantly, the protocol does not specify which patients underwent allograft biopsies. It is clear from the published report that rejection was diagnosed in some patients without a biopsy. In an open label trial such as the current study, whether a clinician chose to perform a biopsy in a specific patient with less certain features of clinical rejection might have been influenced by their knowledge of their induction treatment. This could distort findings and could lead potentially to some patients being misclassified. This problem is likely somewhat minimized since it appears that the majority of participants may have undergone a biopsy at 6 and 12 months.

It remains of concern that patients who received alemtuzumab as induction as compared to patients induced with conventional induction regimens had a significantly higher rate of late biopsy-confirmed acute rejections defined as a biopsy confirmed acute rejection occurring more than 1 year post transplant. The rate was not significant when high-risk and low-risk patients were analyzed separately, but since late rejections were uncommon, the study may not have been sufficiently powered to evaluate this question. The authors also make a point of mentioning that antibody-mediated rejection was not studied. Since such rejection remains a significant problem in transplantation, the incidence of antibody-mediated rejection in patients induced with alemtuzumab versus those induced with a conventional induction regimen is a relevant issue.

Finally, biopsy proven rejection is a surrogate or intermediate outcome and may not adequately reflect patient-centered outcomes. It is quite understandable that the investigators used such an outcome since evaluating more patient-centered outcomes might require 5 or even 10 years of follow-up and delaying some results from this study for 5 or 10 years may be unacceptable. More relevant to clinical practice is whether patients had prolonged preservation of allograft function, whether patients survived long-term, and whether one induction strategy versus another reduced the need for re-transplantation. It would have been helpful had the authors reported the GFR of the trial participants at 6, 12, and 36 months. Follow-up of the patients in this trial at 5 years, 10 years, and beyond might also be informative.

Clinical bottom line: In summary, this large RCT demonstrates that for the primary outcomes of rejection at 6 and 12 months that alemtuzumab is likely equivalent or even superior to conventional induction therapy and not inferior at 36 months by the criteria of preservation of allograft function and patient survival. The issues of antibody-mediated rejection and late acute cellular rejection appear to be unsettled however. In the case of the latter, overall allograft and patient survival did not appear to be adversely effected.