Rethinking the ‘J-Curve’: Study Questions Concern Over Too-Low Diastolic BP

People with a genetic predisposition to low diastolic blood pressure don’t have an excess risk of MI and other cardiovascular events, providing insight into what the “J-curve” seen in prior observational studies actually means.

Analyses using traditional statistical methods did confirm such a relationship, with increased CV risks at both very low and elevated diastolic BP levels, according to a paper published online this week in Circulation. However, mendelian randomization analyses subject to fewer confounders showed that CV risk increased along with a tendency toward higher genetically-determined diastolic readings, without a spike in risk at lower BP levels.

If the J-curve relationship demonstrated in observational studies represented causal harm from low diastolic BP levels, individuals born with a genetic makeup consistent with lower diastolic BP “should be at higher risk for developing adverse outcomes from further lowering of their diastolic blood pressure for environmental reasons, for nongenetic reasons,” John McEvoy, MBBCh (Johns Hopkins University, Baltimore, MD, and National University of Ireland Galway School of Medicine), one of the study’s senior authors, explained to TCTMD.

But that’s not what was observed in the mendelian randomization analyses. “We saw, in fact, that individuals that started off with a low diastolic blood pressure, even as low as 60 [mm Hg], . . . had lower risk for cardiovascular disease and MI over the course of their life,” McEvoy said.

The inference from that, he added, “is that the diastolic blood pressure J-curve may have little to do with diastolic blood pressure itself and more to do with other parameters like vascular stiffness or vascular calcification that are the true culprits in the observation between increased risk and low diastolic blood pressure.”

Concordance With SPRINT

The J-curve concept for diastolic BP and cardiovascular risk first came out of analyses performed in the 1970s in the Framingham Heart Study, and since then there have been hundreds of papers confirming the observational association, McEvoy said. That includes one published by McEvoy’s group in 2016 based on data from the Atherosclerosis Risk in Communities (ARIC) cohort.

A post-hoc analysis of the SPRINT trial, published in 2018, also showed that there was a U-shaped association between diastolic BP and CVD in both arms of the trial. Nevertheless, intensive treatment aimed at lowering systolic BP was beneficial regardless of the level of diastolic BP at baseline. That suggests, McEvoy said, that the J-curve relationship is more likely related to unmeasured confounding or reverse causation than to a causal effect of lowering diastolic BP.

To probe the issue further, he and his colleagues looked at pooled data on 47,407 participants (median age 60 years; 77% women) from five cohorts: ARIC, the Framingham Heart Study, the Cardiovascular Health Study, the Multi-Ethnic Study of Atherosclerosis, and the Women’s Health Initiative. During an average follow-up of 16.5 years, 7.3% of participants had an incident MI.

Applying traditional statistical methods used in observational studies, the investigators confirmed a J-shaped association between diastolic BP and risks of MI and other cardiovascular events, with the lowest risk at a reading of around 70 mm Hg.

More research is always a good thing. John McEvoy

But they then performed two mendelian randomization analyses based on polygenic risk scores reflective of genetically-determined diastolic BP levels. “Mendelian randomization studies exploit the natural randomization provided by the genotypes of each individual to investigate causal relationships, and are therefore known to be minimally affected by confounders,” McEvoy et al explain.

A linear analysis showed that increasing diastolic BP levels were associated with increased risks of CVD outcomes, including MI (HR 1.07 1-mm Hg increase; P < 0.001). A second analysis looking into the possibility of a nonlinear relationship showed no evidence of a J- or U-shaped association between diastolic BP and CV events.

“So all other things being equal, if you’re born with a low diastolic blood pressure on the basis of your genetic makeup, there’s no increase in risk for these events irrespective of what happens down the road as you get older,” McEvoy said, adding that the finding has important clinical implications.

Take, for example, a patient with a high systolic BP but a diastolic reading of about 70 mm Hg, who is at high risk because of the wide pulse pressure and low diastolic BP. “There’s no evidence from SPRINT and now from our mendelian randomization that if you choose to treat that person’s systolic blood pressure to get it to target, and as a consequence further lower their diastolic blood pressure from the baseline, that that will induce harm in that person,” McEvoy said. “If anything, our mendelian randomization data and the SPRINT data would suggest that if you need to get the systolic to target and the diastolic happens to fall as a consequence of treating the systolic that that may in fact be a good thing and not a bad thing.”

A Note of Caution

Commenting for TCTMD, Donna Arnett, PhD (University of Kentucky, Lexington), a past president of the American Heart Association, provided a careful interpretation of the study, pointing out that it cannot provide conclusions about causality.

“Mendelian randomization does not equal causality as people with ‘deadly BP genes’ could have already died and never had the option to take part in the study,” she said in an email. She added, however, that “the diastolic BP polygenic score results suggest that the lower the diastolic blood pressure, the better the outcome, similar to what we have observed for years for measured systolic blood pressure. This aligns with the SPRINT trial, which demonstrated that lowering systolic BP to a lower target (120 mm Hg vs 140 mm Hg) results in better outcomes.”

Still, this new study is not likely to have a major effect on clinical practice, Arnett said. “The authors suggest that these results show that individuals with hypertension but low diastolic blood pressure can be treated aggressively,” she said, adding, however, that “until we have a randomized clinical trial to test that hypothesis, the evidence is still insufficient to make that recommendation.”

McEvoy acknowledged that even with mendelian randomization, the analyses do not reflect causal relationships. “While it gets closer to causality than observational studies, it’s not necessarily true that it represents causality in 100% of cases or that we can be 100% confident that this represents causal effect,” he said.

But the findings align with those from the randomized SPRINT trial, McEvoy noted. “I’m confident these results are compelling and important and have the potential to inform clinical care, but like everything, it’s important to see replication and more research is always a good thing.”