In this study, we conducted a systematic review of RCTs to update our previous meta-analysis of 2004 on the effectiveness of intravenous infusion of protease inhibitors for acute pancreatitis. As the present literature search retrieved no additional RCTs since 2004 up to March 2013, we analyzed a new set of articles including those that had not been identified in the previous meta-analysis . The present analysis re-confirmed the validity of evidence reported to date, as follows: 1) treatment with protease inhibitors did not significantly reduce the mortality rate from acute pancreatitis; 2) subgroup analysis of trials with CMR greater than 0.20 showed limited effectiveness of protease inhibitors in preventing death; 3) protease inhibitors showed no significant effectiveness even in outcomes other than preventing death. Moreover, given that there are no on-going trials regarding the use of protease inhibitors for acute pancreatitis, no new evidence supporting their effectiveness is expected to emerge for the next few years.
In the previous meta-analysis, we evaluated prevention of death due to acute pancreatitis, formation of pancreatic pseudocysts, intra-abdominal abscess, and surgical intervention as outcomes. In the present analysis, we added prevention of abdominal pain due to acute pancreatitis and bowel obstruction. The present analysis had a larger sample size of 1,697 in total, compared to 1,036 in the previous analysis. The effectiveness of protease inhibitors stratified by CMR level was similarly examined in both the previous and present analyses. In the previous analysis , five [34, 43, 44, 47, 49] of 10 selected articles [31, 33, 34, 37, 42–44, 46, 47, 49] were with CMR >0.10 and showed limited effectiveness. In the present analysis, 10 [27, 28, 34–36, 41, 43, 44, 47, 49] of 17 selected articles were with CMR >0.10 and showed no significant effectiveness. On the other hand, the subgroup analysis of articles with CMR >0.20 [34, 49] revealed a significant therapeutic effect. However, it is suspected that low-quality trials that included potentially severe patients might have exaggerated the effectiveness. In fact, the mean Jadad score was 2.5 in RCTs with CMR >0.20, and 1 point in those with CMR >0.30, suggesting that it is very likely that low-quality trials provided a veneer of effectiveness for severe patients.
The quality of evidence was assessed for each trial by two methods, one that used the conventional Jadad scoring system, and the other based on the GRADE system. The mean Jadad score (2.1 points) was low, and the majority of RCTs were deemed low to moderate quality, except for six trials [31, 34, 35, 37, 46, 47] that were deemed high quality with a Jadad score of ≥3. We then used the GRADE system to evaluate each RCT in terms of risk of bias (random allocation, allocation concealment, blinding, outcome data addressed, selective outcome reporting, and others). Furthermore, we graded the quality of the body of evidence for each clinical outcome using the GRADE system in addition to the Jadad scoring system to further strengthen quality of evidence assessment. Handling of outcome data was likely to become a nest of bias, reporting bias, and ITT analysis were mentioned in three articles [44, 46, 47], four articles [36, 41, 47, 49], and one article , respectively. The quality of RCTs included in the present review was low in general, and hence, the bodies of evidence and result estimations are both likely to be low in quality. We also checked each article in terms of whether there were descriptions regarding industrial support, tangible outcome, and sample size calculation in the methodology. With respect to industrial support, we found no significant difference in effectiveness according to the status of industrial support; however, eight articles [28, 32, 33, 36, 44, 46, 47, 49] provided no description regarding funding or medicine support, and thus the actual status was unknown. With respect to tangible outcomes, only four of 17 articles [36, 41, 47, 49] provided descriptions in the methods sections, and no article reported the presence or absence of adverse events. Sample size calculation was described in only one article . The CONSORT statement , which specifies reporting of the primary outcome and sample size calculation, might have led to improved quality of reporting [52, 53], although previous studies found inappropriateness or deficit of reporting common in published articles [54, 55]. It was likely for the authors of adopted trials to select only the preferable results to report, among many outcomes measured.
This study has some limitations worth noting. First, the subgroup analysis which aimed to consider disease severity was not based on established criteria of severity. Instead, we defined CMR >0.10 as an arbitrary and retrospective index to indicate moderate to severe pancreatitis. Prospective indices such as the Atlanta Classification , Ranson’s criteria , or APACHE-II score  were used only in two RCTs [47, 49]. Although CMR is advantageous in that it can be easily calculated after trial completion, clinical interpretation of case severity based on CMR requires prudent consideration, as it is subject to factors such as care quality provided at each institution. Unfortunately, accurate evaluations of the severity of acute pancreatitis are difficult at the entry of each trial. With reluctance, the present study used CMR as a surrogate measure of severity because only two trials [34, 49] had used either the Ranson’s score or APACHE-II score.
The second limitation was the low quality of the individual studies evaluated by our review. The overall mean Jadad score was 2.1 (range, 0–5; three RCTs scored a 0 [28, 32, 43]), which meant that the overall quality of the studies was low to moderate. Our previous meta-analysis  showed that protease inhibitors were effective when CMR was greater than 0.10, while this was true only when CMR was greater than 0.20 in the present study. In our previous report , we calculated the APACHE-II scores using a multiple logistic equation described by Knaus et al. . We also found that CMR = 0.10 scored roughly six points for the APACHE-II score method. On the APACHE-II scale (highest score: 67 points), we discovered that our patients scored 6 points when they presented with high-grade fever, hyperventilation, acidosis, renal dysfunction, leukocytosis, and deterioration of consciousness. One study has already validated the severity of acute pancreatitis with the APACHE-II score system . CMR >0.20 for acute pancreatitis may be worse than these physical conditions, and we would be interested to test whether protease inhibitors would be effective in preventing death or other complications. Although meta-analyses of the two trials with CMR >0.20 showed that protease inhibitors were effective, reporting quality and heterogeneity should be taken into account when interpreting the results [34, 49].
Third, the treatment modality of protease inhibitors was confined to intravenous administration. The present study excluded trials of intra-arterial or intra-abdominal infusion of protease inhibitors because the main objective was to update our previous meta-analysis . Accordingly, we did not evaluate the effectiveness of protease inhibitors according to their various administrations. Further systematic reviews and meta-analyses on these modalities of administrations are expected in future. Future RCTs in this field should examine the effectiveness of protease inhibitors administered through intravenous or intra-arterial infusion, particularly among patients with severe acute pancreatitis.