Acute Inflammatory Markers and Early Stent Thrombosis: Association or Causation?

Stent thrombosis (ST) is a catastrophic event and may occur early (0–30 days), late (30 days to 1 year) or very late (>1 year) after stent implantation. Death after stent thrombosis is high, with an in‐hospital mortality rate of ≈4% for late and very late ST and 8% for early ST and overall mortality rate >20% at 6 months. ¹ , ² In the era of first‐generation drug‐eluting stents, late and very late ST represented the majority of events and contributed to an overall 5‐year ST rate of ≈5% in clinical trials of “real‐world” use. ³ In contemporary clinical trials using latest‐generation drug‐eluting stents, late and very late STs have been reduced to 0.1% to 0.2% per year, while early ST persists at ≈0.5% to 1.0%. ⁴ , ⁵ The large number of stent procedures and the dire consequences of early STs highlight that even this level of risk is a substantial public health concern. The major modifiable risk factors of early STs including untreated dissections, poor stent expansion, and nonadherence to dual antiplatelet therapy are well known. The ongoing hazard suggests that potential other patient‐specific risk factors may be present. To the extent that these can be identified and mitigated, ST and related death could be reduced further.

In this issue of the Journal of the American Heart Association (JAHA), Krychtiuk et al ⁶ present an analysis testing the hypothesis that preprocedural acute inflammatory activation, defined as CRP (C‐reactive protein) >50 mg/L or leukocyte count >12 g/L, is associated with increased risk for early ST. Within a single‐center prospective registry including 11 397 patients undergoing percutaneous coronary intervention, preprocedural inflammatory markers were available for 6880 patients. Overall, early ST occurred in 91 (0.8%) patients and the 6‐month mortality rate after ST was 22%. Notably, there were significant differences in multiple baseline characteristics between the group with inflammatory markers available (the analysis set) and the group in which they were missing, most notable for increased frequency of acute coronary syndrome in the group with available inflammatory markers. Thresholds for leukocyte count and CRP were derived from the internal data on the basis of the calculated C‐index. The high threshold for CRP was selected as to be more specific with acute inflammation rather than chronic low‐grade inflammatory states.

Either CRP >50 mg/L or leukocyte count >12 g/L was present in 24.9% of patients and was associated with increased risk of early ST. The impact on the risk of early ST varied, with greater effect on early ST rates if both markers were elevated and greater relative effect for elevated CRP (3.6 versus 1.1%) than elevated leukocyte count (2.4 versus 1.0%) (Figure). Early ST rates were between 0.8% if neither marker was elevated and 3.9% if both markers were elevated. If CRP was elevated, then there was little difference in early ST rates whether leukocyte count was >12 g/L or not (3.9 versus 3.3%). In contrast, the effect of elevated leukocyte count was greater if CRP was >50 mg/L (3.9 versus 2.1%). Nevertheless, there was a significant univariate association for either marker.

Figure 1. 2×2 table depicting proportion of patients with and without CRP>50 mg/L and leukocyte count >12 g/L and the associated relationship to early ST.

CRP indicates C‐reactive protein; ST, stent thrombosis; and WBC, white blood cell count.

In addition to the univariate association, there are several findings that support the strength of the relationship between the presence of either acute inflammatory marker and early ST. The analysis of multiple subgroups showed homogeneity of the results. Specifically, although acute inflammatory activation was more common among patients presenting with acute coronary syndrome (33.7% versus 16.2%; P<0.00001) than stable presentation, the relative impact of acute inflammatory activation assessed by either inflammatory marker on the risk of early ST was similar or greater for stable presentation (hazard ratio [HR], 3.57 [95% CI, 1.72–7.41]) than for acute coronary syndrome (HR, 2.63 [95% CI, 1.56–4.44]). Likewise, the association of acute inflammatory marker elevation with early ST was similar for relative risk and statistical significance among patients receiving first‐generation (3.0 versus 1.0%; P=0.004) or second‐generation (1.7% versus 0.5%; P<0.001) drug‐eluting stents. Furthermore, the strength of the association was increased by the presence of a dose–response relationship with progressively higher risk of early ST for CRP level and leukocyte count with increasing quartiles of either marker. Finally, the investigators determined a highly significant independent effect of preprocedural acute inflammatory activation after adjusting for other known risk factors while maintaining similar risk estimation and statistical significance (HR, 2.89 [95% CI, 1.86–4.48]; P<0.0001).

The investigators should be commended for the quality of the data from a comprehensive percutaneous coronary intervention (PCI) database and for detailed analysis of a novel hypothesis. There are, however, several limitations that must be considered to place the interpretation of the findings in proper perspective. The data are from a single center, and the thresholds for the acute inflammatory markers were derived and tested in this internal data set without external validation. The ascertainment of the inflammatory markers was incomplete, with significant differences in multiple baseline characteristics, which suggests potential bias for selection of higher‐risk patients in the analysis. Finally, consistent with the observational status of the data, residual confounding is possible. There is some reassurance for the generalizability of the results in that the overall rate of early ST and 6‐month mortality rate were similar to other reports. Also, given the consistency of results across subgroups, it is unlikely that selection of a higher‐risk population for inflammatory marker collection impacted the relative association of elevated markers with early ST.

Allowing for these limitations, what can we infer from the results? Certainly, there is a clear and highly significant association between elevation of leukocyte count or CRP and early ST. The risk is increased nearly 3‐fold even after adjustment for other risk factors. Given this strong association, it is tempting to infer a causative relationship for preprocedural acute inflammatory activity and early ST. Such a relationship would allow planning strategies including possible additional pharmacologic therapies or modifications in timing or technical aspects of the procedure to help reduce subsequent early ST. The authors' suggestion of obtaining CRP routinely and delaying percutaneous coronary intervention in stable patients with elevated CRP or leukocyte count implies a belief in a causal relationship. While such a response may be reasonable for patients with obvious active infection such as pneumonia or sepsis, in whom the risk for early ST was also high, it is still uncertain that the presence of elevated CRP or leukocyte count reflects a causative mechanism or whether the findings of elevated risk associated with the inflammatory markers can be extrapolated to any group of patients for whom the assessment of these markers may guide timing of the procedure, antithrombotic management, or anti‐inflammatory strategies before planned percutaneous coronary intervention.

Unfortunately, the limits of correlation based on observational data prohibit a causal inference regardless of the strength of the association. Additional evidence and analyses are required to move from an associative relationship to causal inference. ⁷ Associations can be inferred on the basis of samples drawn from a distribution using standard statistical analysis so long as the experimental conditions are fixed. Causal inference requires analysis of probabilities when external conditions are modified. These modifications may be based on treatment or other external factors that may be expected to alter the association. Pearl⁷ notes that examples of causal concepts include terms such as randomization, effect, counterfactual analysis, intervention, and instrumental variables, while associations are defined by terms such as correlation, regression, propensity score, and risk ratio.

Given the strength of the association and importance of the outcome, the hypothesis generated from the work of Krychtiuk et al deserves further testing to assess for probability of a causative relationship. First, the associations and optimal threshold of the inflammatory markers identified should be confirmed in a large external data set. If the associations are supported, then studies should be designed to investigate causative relationships and mechanisms. Such studies may include investigations of the impact of delayed percutaneous coronary intervention on levels of inflammatory markers and early ST outcome among stable patients with baseline elevated inflammatory markers, interventional clinical trials of anti‐inflammatory therapies, or clinical trials of more potent P2Y₁₂ inhibitors versus clopidogrel in stable patients with elevated inflammatory markers.

The strength of the associations identified in the current study and the grave importance of early ST notwithstanding, early ST remains an infrequent event, and studies of risk prediction or preventative therapies require large samples and are expensive. Confirmation studies for the association may be conducted within large prospective registries or by combining data from single‐center databases. This should be relatively quick, easy, and inexpensive. There is also much interest and discussion in conducting interventional clinical trials within large prospective registries. The questions posed for the proposed randomized clinical trials above would be well suited for such large, pragmatic designs. A potentially important hypothesis and target for intervention has been identified. Moving from association to possible causation and thereby prevention may allow reduction in a persistent unacceptable risk of coronary stenting.

Disclosures

None.