Are They in the SPRINT Zone? Applying Clinical Trials to Individual Patients

December 12, 2018

The results of randomized trials, with their rigid entry criteria, often don't apply to "the patient in front of you," it is often observed. But perhaps such results could still inform treatment decisions if clinicians knew how similar or different that patient is compared with the trial's "average" participant.

That's the idea behind a proposed scoring system, meant to be a decision aid rather than a decider, that could potentially expand the scope of patients who could be helped by the evidence base for a treatment or strategy, say researchers.

Even if patients are, for example, older than the clinical trial's age cutoff, or have renal function outside its entry-criteria limits, their "trial score" could show that "maybe they're not so different from the patients enrolled in the trial, or maybe they are," Luke J. Laffin, MD, Cleveland Clinic Foundation, told theheart.org | Medscape Cardiology. "It gives physicians another tool to judge that."

Laffin is lead author on the analysis published online November 26 in the European Journal of Preventive Cardiology.

A Composite of Six Variables

He and his colleagues used the SPRINT trial as a test case for developing the scoring system, which they calibrated with help from 2007 to 2014 NHANES participants without diabetes as a proxy for the broad clinical-practice population.

SPRINT, which compared two strategies for lowering systolic blood pressure (BP) in patients with cardiovascular (CV) risk factors but no diabetes, had a profound impact on recent hypertension guidelines. It saw significant reductions in risks for death and CV events in patients treated to a systolic BP of less than 120 mm Hg compared with a target below 140 mm Hg, which was consistent with the guidelines of the time.

Patients' scores represented a composite of their differences from the means for six individual continuous-variable features of SPRINT participants at baseline: age, systolic BP, fasting serum glucose, non-HDL-lipoprotein cholesterol, serum creatinine, and body mass index.

The lower the composite score, the shorter the patient's "distance" from a SPRINT patient as averaged over those six variables; the lower the score, the more resemblance to those in SPRINT.

It has been common, Laffin observed, for clinicians to conclude that SPRINT doesn't apply to one or another of their patients embarking on antihypertensive drug therapy. "SPRINT was a perfect trial for this, because it looked at one straightforward question that was really practice-changing in a lot of respects, but enrolled a relatively narrow population, when you look at the inclusion criteria."

"But What About the Flip Side?"

This approach to a common quandary "is at least a step in the right direction," said Sripal Bangalore, MD, MHA, NYU School of Medicine, New York City, "but I don't think it's as simple as that. How do you choose your baseline variables, and what about unmeasured variables that we look at in clinical practice?"

For example, a metric reflecting frailty would be useful, Bangalore, not connected with the current study, told theheart.org | Medscape Cardiology.

And, does the trial-score system add anything to established metrics, such as the DAPT Risk Score, he asked. Also, the trial score focuses on benefit, "but what about the flip side? It really doesn't look into the risk–benefit ratio for these patients."

For the current analysis, trial scores were calculated for the nondiabetic NHANES cohort with systolic BP above 130 mm Hg to determine their distribution, from which three "data zones" were defined: data-rich, that is below the 90th percentile; data-limited, from the 90th to 97.5th percentile; and data-free, above the 97.5th percentile.

For any patient with a SPRINT trial score falling into the data-rich zone, "I think what we can say is that they're similar to a patient enrolled in SPRINT, and so would likely benefit from an aggressive blood-pressure treatment strategy," Laffin said.

"What we're saying about the data-limited and data-free zones is not necessarily that those patients won't benefit from a SPRINT strategy, that is, more aggressive blood-pressure control. We're saying that those patients are not well-represented by SPRINT," he said. "SPRINT shouldn't be the trial that informs us about those patients."

Interestingly, a SPRINT patient who was average for all six parameters would, by definition, have had a trial score of zero, and that patient didn't exist.

The scoring system was applied to patients of the ACCORD-BP trial, who were hypertensive but also, unlike SPRINT patients, had diabetes at baseline. That was thought to make them different enough to plumb whether the scores worked well.

ACCORD-BP also compared aggressive systolic BP control with a standard target (<120 vs <140 mm Hg), and used end points similar to those in SPRINT. Famously, there was no significant difference in that trial between the two strategies for the primary clinical end point.

Only 28% of ACCORD-BP patients had scores suggesting they were similar to the average SPRINT patient. The corresponding number for the nondiabetic NHANES cohort was about 70%, the group found.

For perspective, a 2016 analysis showed that 16.7% of patients receiving treatment for hypertension in a 2007 to 2012 NHANES cohort, which included diabetic patients, would have been eligible for SPRINT on the basis of its entry criteria.

Striking it Data-Rich

The composite clinical outcome risk was significantly lower for SPRINT patients than for the majority of ACCORD-BP patients, with scores putting them in the SPRINT data-free or data-limited zones, regardless of whether they were treated to the aggressive or standard systolic BP target.

Hazard Ratio (HR) for Composite Clinical End Point* by ACCORD-BP Patient-Score Data Zones and Systolic-BP Target Group
Comparison HR (95%CI) for Intensive BP Treatment Group HR (95%CI) for Standard BP Treatment Group
SPRINT vs data-rich ACCORD-BP patients 0.97 (0.69–1.35), P = .84 0.95 (0.72–1.25), P = .70
SPRINT vs data-limited/free ACCORD-BP patients 0.64 (0.52–0.78), P < .01 0.77 (0.64–0.93), P < .01
*SPRINT primary end point of MI, non-MI acute coronary syndrome, stroke, acute decompensated heart failure, or CV death. Heart-failure events were a secondary end point in ACCORD-BP that was combined with its primary end point to make it consistent with and allow comparison to SPRINT.

Conversely, the outcomes risks for SPRINT patients and for ACCORD-BP patients scoring in the SPRINT data-rich zone, regardless of randomized systolic BP target, were statistically similar.

Not every guidelines-based treatment or strategy is based primarily on a single trial, at least to the extent SPRINT has influenced the management of hypertension, so the trial score system as outlined in the current report might not work the same way across cardiology.

But Laffin offered examples of several other randomized studies with data-rich zones that might be useful for selecting patients likely to respond to the tested treatment or strategy as observed in the trial.

They include the PARADIGM-HF trial testing sacubitril/valsartan (Entresto, Novartis) in patients with chronic heart failure, he said, and the DANISH trial, which questioned the effectiveness of implantable cardioverter defibrillators in patients with nonischemic cardiomyopathy.

Laffin discloses receiving speaking fees from EP Consulting; disclosure for the other authors are in the report. Bangalore has recently disclosed serving on advisory boards for Abbott Vascular, Biotronik, Amgen, Pfizer; and receiving research grants from Abbott Vascular and honoraria from Abbott Vascular, Pfizer, AstraZeneca, and Biotronik

Eur J Prev Cardiol. Published online November26, 2018. Abstract

J Am Coll Cardiol. 2016;67:463-472. Full text

Follow Steve Stiles on Twitter: @SteveStiles2. For more from theheart.org | Medscape Cardiology, follow us on Twitter and Facebook.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....