Text version
Do we finally have a lab test that can guide your choice of antidepressants based on more than just estimated cytochrome metabolic rates? Here’s a new report suggesting that serotonin–norepinephrine reuptake inhibitors, like venlafaxine, work better than a serotonin reuptake inhibitor, like sertraline, in patients with a high C-reactive protein (CRP) level. But don’t change your practice just yet. Let’s have a look at this study and see whether its findings are strong enough to make you order a CRP level before shared decision making with your patient about which antidepressant to consider.
Hi! Jim Phelps here for the Psychopharmacology Institute. First, do enough patients have high CRP levels to justify even looking for them? Well, yes. In this study of 918 patients, 23% had a level over 1 mg/L. Okay. Next step. How was “works better” defined? Oh darn. They used the Clinical Global Improvement (CGI) scale—not a very robust instrument. But before I go poking any more holes in this study, CRP measures inflammation, right? And inflammation is clearly connected to depression. So, in the long run, this notion of CRP as an easy to measure indicator is very appealing. Indeed, this new article is not the first to suggest that CRP could be a useful biomarker. Remember the 2014 randomized trial in which patients with high CRP had better treatment outcomes with nortriptyline than with escitalopram? By contrast, this new study is simply a retrospective examination of patient records. As such, the authors had to control statistically for potential confounding factors, like age, gender, anxiety and other comorbidities, such as obesity, which is associated with high CRP levels. You will see that this statistical correction is crucial in interpreting the results.
The main finding. In patients with a CRP less than 1, outcomes were the same whether patients were treated with an SRI or an SNRI. But as reported in the abstract and in a prominent psychiatry newsletter, patients with a CRP of 1 or higher had better outcomes with SNRIs than with SSRIs. This was true only after controlling for those confounding factors. That was done using the Cox proportional hazards regression model—the details of which are still beyond me, which is unfortunate because this result has huge clinical implications. Should we all start measuring CRP levels and recommend SNRIs rather than SRIs for patients with a CRP of 1 or higher? On the other hand, you don’t have to understand Cox regressions to recognize the difference between a randomized trial and a retrospective cohort study. Likewise, you don’t have to be a research specialist to understand that the CGI, which was the outcome measure used here, calls for subjective judgments of improvement based on chart review, not direct patient interview as with the Hamilton or Montgomery-Asberg rating scale.
And finally, we still have to ask if the results were statistically and clinically significant. The p-value for the advantage of SNRIs over SSRIs in patients with a CRP of 1 or more was 0.037. Clinically significant? For that, I think you have to look at Figure 2B of the article, which is linked here at the Psychopharmacology Institute. It sure looks like a small difference to me, but you might detect some bias in my view. Why? Because I spend an awful lot of time tapering people off of antidepressants, and venlafaxine, as you have likely discovered, is among the hardest to discontinue. In my view, even with a robust difference in the outcome, this study would still be insufficient to justify using SNRIs based on CRP because of its retrospective design and weak outcome measure.
A better study on which to hang clinical decision making would be the 2014 randomized trial in which nortriptyline was favored over escitalopram in patients with high CRP levels. But that was nortriptyline, not venlafaxine, and the study was 12 weeks long. So, like all of our short-term efficacy studies, it doesn’t give us insight into discontinuation withdrawal. Nevertheless, as the authors of this new paper review in their discussion, multiple studies have suggested that high CRP levels may predict some resistance to serotonergic antidepressants. Likewise, obesity clearly increases the likelihood of having a high CRP.
So, before turning to venlafaxine or nortriptyline for patients with a high CRP, wouldn’t logic suggest bupropion? That reminds me of a commentary from 2005, perhaps still relevant today and maybe even more so in the context of CRP levels. It was titled, “Why isn’t Bupropion the Most Frequently Prescribed Antidepressant?” Answer: Clinicians believe bupropion can make anxiety worse. A 2008 review cited in that commentary does not support this belief. For more, see the references.
For now, to summarize, although the headlines and the abstract of this new retrospective study suggest that clinicians should obtain CRP levels and treat patients with a result of 1 or higher with an SNRI, the data are not sufficiently strong to justify turning to higher risk antidepressants on this basis alone. Nevertheless, watch for more on this issue because inflammation matters, and perhaps CRP can be a guide somehow.
Abstract
C-Reactive Protein Could Predict the Efficacy of SSRIs in Clinical Practice: A Cohort Study of Large Samples in the Real World
Yuqian Pan, Rui Luo, Shuqi Zhang, Yuxia Liu, Yiping Wang, Simeng Feng, Hengfen Li
Background: C-reactive protein (CRP) has been shown to predict antidepressant treatment outcomes in several trials, but they were limited to small-sample and strictly-restricted conditions. This study plans to verify if CRP can predict antidepressant efficacy in large samples in the real world.
Methods: 918 depressed patients who had tested CRP were included, then were followed up through their outpatient visits or by telephone to obtain information about their medication therapy (SSRIs, SNRIs, MT, NaSSA) and assess efficacy using the Clinical Global Impressions-Improvement scale (CGII). Efficacy was classified as effective and ineffective and CRP was separated into the low CRP group (CRP <1 mg/L, n = 709) and the high CRP group (CRP ≥1 mg/L, n = 209).The efficacy was compared in different groups.
Results: Using Kaplan-Meier survival analysis and Cox proportional regression model to analyze, it was discovered that SNRIs were more effective than SSRIs in treating patients with high CRP(HR = 1.652, p = 0.037,95 % CI:1.031-2.654), and SSRIs were more effective in treating patients with low CRP than those with high CRP (HR = 1.257, p = 0.047,95 % CI:1.003-1.574), while no difference in efficacy between the two groups was found in patients using SNRIs, MT, NaSSA.
Limitations: Small amounts of MT and NaSSA were included, and some factors that may affect CRP value have not been controlled.
Conclusion: CRP could predict the efficacy of SSRIs in the real world, depressed patients with high CRP may be more likely to respond poorly to SSRIs.
Download PDF and other files
Reference
Pan, Y., Luo, R., Zhang, S., Liu, Y., Wang, Y., Feng, S., & Li, H. (2022). C-reactive protein could predict the efficacy of SSRIs in clinical practice: A cohort study of large samples in the real world. Journal of Affective Disorders, 313, 251-259.
