Sindhuri Prakash-Polet, MD, PhD & Sean Kalloo, MD, MBA, FASDIN
Fellow in Interventional Nephrology & Assistant Professor of Medicine (in Radiology), Columbia University Medical Center
Introduction: Efforts to find therapeutics and interventions that can modify the trajectory of chronic kidney disease and potentially delay the progression of kidney failure are ongoing. One such intervention has been getting some press lately. Observational data have shown that the very creation of the arteriovenous fistula in preparation for dialysis is associated with delayed progression to kidney failure and initiation of renal replacement therapy. This blog post examines the data behind this observation and if creation of arteriovenous fistula is an incentive to alter chronic kidney disease progression.
Does creation of arteriovenous fistula delay progression? In 2015, Golper et al observed that the creation of arteriovenous fistula in a small heterogenous sample of 123 patients from a single center can delay the initiation of renal replacement therapy. They determine this trend by comparing mean eGFR during an observation period of 2 years before and after fistula creation over a total period of about 4 years. They note that eGFR declined from 28.3 ml/min to 16.9 ml/min before fistula creation but declined only to a mean of 15.8 ml/min after creation. Follow up of creatinine stopped if the patient initiated dialysis, succumbed to death or was transplanted. It is unclear what the mean follow up period was in these 3 subgroups. At the outset,It is difficult to accept the conclusions drawn in this study due to several limitations. All patients who received an arteriovenous fistula were assumed to be due to progression of chronic kidney disease. Patient characteristics such as age, sex, medication use, presence of co-morbid conditions were not used in the analysis. More importantly, eGFR was calculated using the MDRD equation, which performs better than CKD-EPI equation in patients with lower eGFR but patient weight over the time period was not considered which can affect creatinine. Additionally, albuminuria was not documented which is correlated with progression of kidney disease. In addition, patients were not compared to any control group in this study nullifying the conclusions. During the 2 year follow up period after creation of arteriovenous fistula the majority of them (~ 58% (n=72) initiated dialysis. Only about 30% did not initiate renal replacement therapy and 5% were transplanted and might represent “regressors.” Also, this cohort had a significantly lower mortality rate (about 0.03%).
A follow up single center study using patients with advanced kidney disease compared eGFR trajectory between patients who had arteriovenous fistula and patients who underwent peritoneal dialysis catheter placement. A total of 61 patient pairs were compared and demographic and clinical data including albuminuria were included in the analysis. Duration of follow up was about 10 months before and after fistula creation and catheter placement and stopped if the patient died, initiated dialysis, or underwent a kidney transplant. Patients undergoing peritoneal dialysis initiated dialysis more often than those planned to undergo hemodialysis (79% vs 39%). Rate of eGFR (ml/min per 1.73m2) decline was -2.5 after fistula creation from -4.1 and -4.5 from -5.3 after peritoneal dialysis catheter placement. Using mixed effects linear regression, predicted annual decline decreased from -5.1 to -2.8 after fistula creation when compared to -5.5 to 5.1 per year in the patients with peritoneal dialysis catheter. The attenuation of eGFR was statistically significant in both crude eGFR measure and predicted eGFR measure accounting for sparing of 0.14 of eGFR decline with fistula creation. These results were similar after adjustment for age, sex, use of reno-protective agents and diuretic use. Patients who underwent fistula creation had a higher BMI on average when compared to patients who planned to undergo peritoneal dialysis, and patients who underwent catheter placement had lower eGFRs when compared to hemodialysis suggesting that they were more ill at the time of analysis. Nonetheless it is difficult to ascertain a causal relationship in a retrospective study. Fistula creation occurs earlier than PD catheter placement, which might introduce some performance bias and allow patients to employ lifestyle modifications that can delay progression. Moreover, given that data was collected at a single center with a small sample size the generalizability of the data is also in question.
Similar conclusions were drawn from a larger retrospective study with about 3000 United States veterans comparing patients with arteriovenous fistula or graft and patients who initiated dialysis with a catheter. Specifically, they compared eGFR trajectory 6 months prior to dialysis initiation with catheter vs the 6 month period after arteriovenous fistula or graft creation. The authors found that there was a deceleration of eGFR after vascular access creation that was independent of maturation. Again, given the retrospective nature of the study, causality cannot be inferred. In addition, patients who initiated dialysis with the catheter appeared to be sicker although the authors adjusted for confounding variables.
Is there a physiologic basis to this observation? The authors argue that the creation of arteriovenous fistula involves repeated clamping of the feeding artery that might lead to remote ischemic preconditioning and explain the observed phenomenon. In addition, fistula creation predisposes to distal limb ischemia allowing sustained or periodic bouts of ischemia. Remote ischemic preconditioning involves inducing transient ischemia and reperfusion before a surgery or procedure that would cause prolonged ischemia to the organ in question. This would theoretically protect other target organs in distant sites by upregulating various humoral, neuronal and systemic pathways. Initially described in cardiac tissue, where a larger area of the myocardium was salvaged after ST-elevation myocardial infarction in patients undergoing remote ischemic preconditioning, other target organs seemed to benefit as well. Similarly, patients who underwent remote ischemic preconditioning prior to cardiac surgery had reduced rates of acute kidney injury, with fewer patients receiving renal replacement therapy. The procedure involved 3 cycles of 5-minute inflation and 5-minute deflation using a blood pressure cuff in the upper arm prior to the surgery. This is thought to occur by inducing molecules that induce cell cycle arrest early in the kidney protecting them from further injury during cardiac surgery. Similarly several other studies show reduction in acute kidney injury after remote ischemic preconditioning in healthy individuals and patients with moderate kidney impairment (eGFR <60 ml/min/ 1.73m2). However, conflicting results have emerged in studies that show no effect, thus halting adoption to clinical practice. In addition, the mechanism by which renoprotection occurs appears to be in the context of acute kidney injury which is more pronounced in patients with underlying chronic kidney disease, but studies interrogating remote ischemic preconditioning on disease progression in patients with advanced chronic kidney disease (eGFR < 29ml/min/1.73m2 ) is lacking. In a study that examined remote ischemic preconditioning on progression of renal impairment after partial nephrectomy, eGFR reduction was lower after 1 month when compared to controls but failed to sustain non-progression after 6 months. Thus at present, the effect of remote ischemic preconditioning on advanced kidney disease and disease progression is yet to be determined.
A plausible explanation for reduction in kidney disease progression is the effect of arteriovenous fistula on renal hemodynamics. Given that the arteriovenous fistula is a low resistance vessel, overall systemic vascular resistance will be reduced evidenced by reduction in systolic and diastolic blood pressures. In addition, the shunt increases preload augmenting pulmonary flow and cardiac output leading to increased oxygen delivery to the distal tissues and reduction in renal ischemia. These concerted effects might delay progression of kidney disease after fistula creation.
Conclusion: The finding that creation of arteriovenous fistula delays kidney disease progression is yet to be investigated using prospective clinical trials. How much of the observed phenomenon is patients who will ultimately regress given the natural history of advanced kidney disease needs to be deciphered. Perhaps investigating whether this is also observed in the subset of patients who are at high risk of progression such as patients who are older and with severe albuminuria might be the next step in determining an association or causality.
Reviewed by Matthew A. Sparks, Edgar Lerma
Addendum (10/15): We added this addendum to clarify some of the points of the post after discussion on Twitter. The observational data likely represents the natural history of chronic kidney disease patients where a large portion of patients experience regression to the mean and thus do not need to initiate kidney replacement therapy. Moreover, it was pointed out that the creation of arteriovenous fistula can help with patient engagement by motivating lifestyle modifications but such studies should not advocate for fistula creation as access type should be individualized for every patient. Thus, the above observational studies have significant methodological flaws. It is unlikely that the AVF creation improved kidney function for the aforementioned issues. However, it is also worth noting that the AVF creation does not lead to worsening of kidney function.