Midodrine Is an Effective Therapy for Resistant Intradialytic Hypotension: PRO

Intradialytic hypotension (IDH) presents a significant clinical challenge in the management of hemodialysis patients. The medical literature contains numerous definitions of IDH; hence, the exact effect is hard to pinpoint, ranging from 8% to 40% of treatments.1 Investigators have identified at least 15 different published definitions, including those based on nadir systolic or mean arterial pressures below a particular value or of a certain magnitude, presence of symptoms, or management interventions.2 Nonetheless, IDH has a strong association with poor outcomes;2–4 hence, strategies have been proposed to manage this, with varying success and levels of evidence.1 As recently reviewed,1 IDH results from the interplay between ultrafiltration, cardiac output, and vascular tone. As fluid is removed during treatment, cardiac output may fall as a failure of compensation of heart rate, myocardial contractility, vascular smooth muscle tone, redirection of splanchnic flow, and venous capacitance and plasma refilling. Osmotic shifts related to solute removal also contribute to contraction of the extracellular fluid volume. Furthermore, IDH contributes to myocardial stunning and impaired myocardial contractility,5 which begets further IDH in a vicious cycle. The threat to myocardial perfusion during hemodialysis is mirrored in many other organ systems, including the brain, liver, gut, and kidneys6 and likely underpins the heightened mortality and morbidity, cognitive decline, uremic symptoms, and poor quality of life experienced by many patients. Strategies to prevent or treat IDH include nonpharmacologic approaches and dialysis prescription changes.7 For instance, cooling dialysate will lead to peripheral vasoconstriction and support of BP. In a small, randomized, crossover study, this led to fewer regional wall motion abnormalities (an echocardiographic measure of myocardial stunning) and improved systolic pressure.8 When implementation of these strategies fails to ameliorate IDH, one of the most widely used pharmacologic approaches is midodrine. Midodrine, acting through its metabolite, desglymidodrine, is an oral adrenergic α1-agonist that has been used for many decades in the management of orthostatic hypotension in autonomic neuropathy and spinal cord injury.9 Midodrine is rapidly and almost completely absorbed from the gastrointestinal tract. Enzymatic cleavage of a glycine group to the active metabolite occurs promptly, leading to a peak of desglymidodrine 1–1.5 hours after ingestion. Excretion of desglymidodrine occurs primarily through the kidneys, hence the 3-hour half-life in normal renal function is extended to 9–10 hours in ESKD. However, both midodrine and desglymidodrine exhibit low protein binding, and are relatively small molecules. So, the half-life during dialysis is only around 2 hours.9–11 The effects on BP arise from both arterial and venous constriction. The former leads to increased peripheral vascular resistance, whereas the latter allows for rapid return of blood from capacitance vessels, including splanchnic veins, increasing ventricular filling and improving cardiac output. Midodrine is effective in the prevention and treatment of IDH. In a meta-analysis12 including 117 patients from 10 studies, postdialysis systolic BP in the treatment group was higher by 12.4 mm Hg (95% confidence interval [95% CI], 7.1 to 17.7), and diastolic BP was higher by 7.3 mm Hg (95% CI, 3.7 to 10.9). Likewise, the nadir systolic BP was higher by 13.3 mm Hg (95% CI, 8.6 to 18.0), with a difference in nadir diastolic BP of 5.9 mm Hg (95% CI, 2.7 to 9.1). Most studies reported improvement in IDH symptoms, and there were no reported serious adverse events, although the studies were small, and follow-up was brief. In terms of the effect on clinical outcomes, data are very limited. In a study of 50 chronically hypotensive nondialysis patients, midodrine was found to improve middle cerebral arterial blood flow as measured by transcranial Doppler and improve attentional performance in cognitive testing.13 These results were similar to a study of hemodialysis patients with chronic orthostatic hypotension, defined as a drop in standing postdialysis systolic BP of >20 mm Hg, whose cerebral blood flow was preserved during tilt-table testing with the administration of midodrine.14 The lack of long-term safety data from rigorous randomized trials cannot be ignored. In a retrospective observational study of 1046 patients who were prescribed midodrine for IDH and 2037 matched controls, midodrine use was associated with higher adjusted incidence rate ratio of mortality (1.37; 95% CI, 1.15 to 1.62), all-cause hospitalization (1.31; 95% CI, 1.19 to 1.43), and cardiovascular hospitalization (1.41; 95% CI, 1.17 to 1.71) (15). As this study represents the largest analysis of midodrine for IDH to date, and suggests a safety signal, it bears closer scrutiny. Midodrine patients were very different from the population of controls in aggregate. So, this required careful matching on variables of dialysis vintage, predialysis BP, nadir BP, and percentage of treatments complicated by IDH. These variables were assessed, and hence matched, for the calendar month immediately preceding the month in which midodrine was initiated in the exposed patients. Yet, by the index date of midodrine start, the predialysis and nadir BP were already statistically lower in the midodrine group, and the percentage of treatments with IDH had risen. To mitigate these differences, patients were only included in this analysis when the index date fell within the first 10 days of the month. This now led to insignificant differences between midodrine and control patients at the index date. The obvious question is what led clinicians to prescribe midodrine in the exposed patients, also known as confounding by indication, and whether the statistical maneuvers applied were adequate to protect against residual confounding. When assessing an observational study such as this for confounding by indication, one must consider why clinicians would select specific interventions such as midodrine, and how those clinical decisions might be influenced by factors that are also related to the outcomes of interest. As described by Kyriacou and Lewis,16 patients with severe illness are likely to receive more intensive treatments and, when comparing interventions, the more intensive intervention will appear to result in poorer outcomes. These authors refer to this as “confounding by severity” to emphasize that the degree of illness is the confounder. Because severity of illness affects both treatment selection and patient outcome and is not necessarily an intermediate between the treatment and outcome, it fulfills the criteria for confounding. Risk factors for IDH are, not surprisingly, risk factors for cardiac disease and hence mortality. These include diabetes mellitus, systolic and diastolic dysfunction, ischemic heart disease, arrhythmias, vascular calcification, autonomic dysfunction, age, body mass index, and predialysis chronic hypotension.1 Clinicians who have exhausted all nonpharmacologic means to treat IDH in this population, and then choose midodrine as a next step, are treating a population of patients who are clearly at high risk of the very outcomes demonstrated in this study. Furthermore, severity of illness likely influences treatment decisions leading to downtitration or cessation of therapies entwined within the illness spectrum (e.g., β -blockers for atrial fibrillation and/or heart failure, renin-angiotensin-aldosterone blockade for heart failure, other antihypertensive or -ischemic therapies). The statistical approaches in this study try to account for this heightened risk, but residual confounding may well persist in these types of retrospective, nonrandomized, observational analyses. Do we have other examples of such confounding by indication potentially leading to counterintuitive (and perhaps spurious) results in the dialysis population? In a large, retrospective observational study of heparin use in chronic hemodialysis treatments, investigators identified that patients routinely exposed to heparin had a lower rate of hospitalization for gastrointestinal bleeding.17 In the absence of either a systemic anticoagulant or an antiplatelet agent, the crude rate of hospitalization for gastrointestinal bleed for those not receiving heparin was more than double that of those who were. These authors suggested that one explanation for this finding was “confounding by indication,” given that heparin may be withheld from patients perceived to be at higher risk for bleeding events, and that analytic methods to control for this confounding may not have proved adequate. Another example was a post hoc analysis of the HEMO study where investigators examined the nonrandomized exposure of angiotensin converting enzyme (ACE) inhibitor use, using proportional hazards regression and a propensity score analysis to compare exposed and unexposed patients.18 ACE inhibitor use was associated with a higher risk of hospitalization for heart failure (hazard ratio 1.41; 95% CI, 1.11 to 1.8). In the propensity score–matched cohort, ACE inhibitors were no longer significantly associated with heart failure hospitalization, although the point estimate for the hazard ratio was essentially unchanged at 1.43, with larger confidence intervals. Is one left to conclude that ACE inhibitors are causally responsible for increased hospitalizations for heart failure in dialysis patients? Or is this another example of confounding by indication where, as the authors suggest in their discussion, the results could have arisen from residual confounding not addressed by multivariable adjustment? In the spirit of debate, it is tempting to take a firm position on a topic (in this case, arguing the PRO side of the midodrine debate), ignoring or brushing over weaknesses in one’s own argument while at the same time diminishing the points made by the opposing side. Although the evidence for benefit of midodrine relies on small studies showing improved surrogate outcomes (BP), the retrospective observational study examining for worrisome safety signals was subject to confounding by indication. In lieu of large, randomized trials, we would argue first for studies that examine more physiologically important surrogates than simple BP, for instance focusing on the microcirculation and function of the brain and heart during hemodialysis, and whether midodrine can support tissue perfusion and function and not just crude hemodynamics. A signal of benefit from such studies, which could start with small animal studies and progress to carefully conducted human studies, would justify the time and expense needed to then carry out an adequately powered, prospective, and randomized trial, free of bias and inclusive of clinically meaningful outcomes. A signal of harm would lend support to the CON side of the argument and steer us all in a new direction. Perhaps this is the middle ground to which we can all agree.