Studies have shown that PCT can effectively diagnose infectious complication among solid organ transplant recipients[30–32]. Unfortunately, these studies were sporadic and there were few analyses focusing on single organ, liver, for example. Worse still, drawbacks such as certain methodological flaw, small sample size and lack of united cut-off value existed in these published articles. In this study, we managed to perform a random-effects model analysis on 7 clinical trials from 6 countries to investigate PCT’s utility in differentiating infection among LT recipients. To the best of our knowledge, there have been no previous evidence on this field of concern.
Our results presented a pooled DOR of 18.65 (95%CI 9.85–35.31), suggesting a good diagnostic accuracy for PCT, which was further confirmed by an AUC measuring 0.857, strengthening its reliability for diagnosing infectious complications following LT. Apart from demonstrating the diagnostic value, more importantly, the article also reveals several meaningful but not well understood aspects of PCT’s utility.
Before performing the meta-analysis, PCT concentration, a continuous variable whose cutoff varied among studies, was considered as a potential resource for heterogeneity. Surprisingly, Spearman correlation returned by the Meta-DiSc suggested that no significant threshold effect existed, which meant that different thresholds of various studies were not the main causes for heterogeneity. One possible explanation is that the methods to detect PCT concentration in the included studies somehow neutralized the effect due to the different cutoff settings, leaving an evenly balanced result that passed the threshold effect test which produced a Spearman correlation measuring 0.429 with a p-value of 0.337. Unfortunately, the true reason of it remains unknown and therefore wide discussion as well as more studies is needed. Of note, we found the cutoff value set in the early years are relatively high, which gradually fell down at 0.5 ng/ml in the recent years which presented a high DOR as well as the AUC. Hence researchers are advised to adopt this standard and to verify it in the following studies.
Sensitivity analysis showed that the research conducted by Hara et al led a huge heterogeneity, by deleting which, the I2 was reduced to 0%. The reason accounting for it may be that it’s the only one study evaluated the PCT value as a predictive biomarker. Hara sampled the patients on the day before transplantation while others sampled on the day of diagnosis of infectious disease. It’s a retrospective cohort of 136 liver transplant patients which demonstrated that, to some degree, sepsis can be predicted by pretransplant serum PCT level, for which the authors suggested a re-evaluation of the general condition and rescheduling of LT of a patient whose pretransplant serum PCT > 0.5 ng/ml. Though this study was not included into the meta-analysis due to heterogeneity, it still remains a positive direction to explore.
However, a note of caution is due here since some limitation existed. Due to the inherent flaws in study design, this meta-analysis can only evaluate and combine the size effects measuring the diagnostic accuracy instead of the clinical benefits. Therefore, we call for more studies using random control design with double-blind method to provide solid evidence so as to evaluate the clinical effects from which LT patients may benefit. Besides, as shown in Fig. 3, certain studies have high risk in index test in terms of risk of bias. It is because that in the early years, the universal cut-off was not settled for reasons like different testing methods and incomplete understanding of PCT, resulting in the wide usage of receiver operating characteristic curve for generating an optimal cut-off, which has been overcome in the recent years. Another limitation in this article was that the amount of the included studies is very limited, lowering the level of evidence.