Serum matrix metalloproteinase-7 (MMP-7): As good as it gets?

An expeditious diagnosis of biliary atresia (BA) allowing earlier specific surgical referral would be a good thing, of that there should be no doubt. Delays in diagnosis, perhaps surprisingly, are real issues in North America. The most recently reported series from a consortium of hospitals in the Western United States suggested a mean (SD) of 64 (±25) days,1 compared with our own UK national series of median (IQR) of 48 (35–57) days over a similar period.2 A number of solutions have been suggested including screening newborns for the disease, but that seems a prospect for the future rather than the present.3 An abbreviated diagnostic process in actual infants with jaundice would also help and the real question is—can measurement of serum matrix metalloproteinase-7 (MMP-7) deliver the goods? Let's look back at its origins, as it is illustrative of the shifting nature of BA science over the recent past. It was originally identified as having some relationship in 2 small studies on liver protease expression from Taiwan in the early 2000s.4,5 It then re-emerged as the best discriminatory protein from a large proteomics study using the North American ChiLDReN biobank, published in 2017.6 Its purpose was to identify new diagnostic proteins to try and distinguish BA from any other cause of infantile cholestatic jaundice. Thereafter, a whole host of Asian (mostly Chinese) centers reported their experience, initially with fairly small numbers of patients with BA but later much bigger series emerged (eg, Shanghai, n=1877). All these Asian series strongly inferred that the estimation of serum MMP-7 was almost perfect as a discriminator with, invariably, AUROCs of >0.95. Our own study, from Kings College Hospital, London, UK was retrospective, used biobank samples with 32 infants with BA and 27 cholestatic controls, and was much less discriminatory than the Asian series with an AUROC of 0.83 (sensitivity: 84%, specificity: 78%).7 The current study (BA, n=201), again using ChiLDRen stored biobank serum, now back in North America, is somewhat in between at 0.90 (CI: 0.87–0.94).8 There have been some disconcerting methodological observations in that the cutoff levels varied considerably between studies. This appeared during the wave of enthusiastic Asian studies and was felt to be due to the actual choice of assay. The current study actually reports 2 further assays not previously reported (Luminex assay, MilliporeSigma and Time-Resolved Fluorescence Energy Transfer [TR-FRET] for MMP-7 assay [BioAuxilium]). So on the same patients, we have 2 different cutoffs—52.8 and 18.2 ng/mL, respectively though with similar AUROCs of 0.90 (sensitivity: 94%, specificity: 78%) and 0.88 (sensitivity: 94%, specificity: 66%). Going forward, this will surely limit its applicability in the real world unless there are moves to standardize the assay or one emerges as dominant in the market. Why might there be such a difference between North America and Europe on the one hand and China and Japan on the other? One issue might be that BA is not actually a uniform disease with uniform pathology or indeed, etiology. It is much more common in Asian countries (5-10 per 10,000 births) compared to both Europe and North America (1 in 16 per 20,000) with the former having a greater proportion of viral-associated BA (principally cytomegalovirus) and a much lower proportion of syndromic variants. Such tests invariably assume uniformity of disease and natural history and the rarity of the disease limits any deeper exploration. Is MMP-7 reflecting something intrinsic and specific in BA? Or, is it simply related to the observation that BA, unlike the typical diseases one is trying to discriminate from (idiopathic neonatal hepatitis makes up 58% of the controls here, for instance), has a much more aggressive accompanying fibrotic process? MMP-7 is secreted by Kupffer cells/macrophages, biliary epithelial cells, and fibroblasts and has a proteolytic role in the extracellular matrix of the liver. When first identified, it was felt to be specific to BA, and more specifically the degree of fibrosis.4,5 Nevertheless, our study9 and others7,10,11 have shown only a modest correlation of serum MMP-7 with liver fibrosis at the time of Kasai portoenterostomy. The false positives among the cholestatic controls in the current study (n=44, 22%) were certainly older and had significantly higher liver enzyme profiles (gamma-glutamyl transferase, alanine aminotransferase, aspartate aminotransferase). Whereas nothing clinically significant emerged in those deemed falsely negative (n=12, 6% of infants with BA). Interestingly, other older biomarkers were tested on the same specimens by way of comparison. Gamma-glutamyl transferase had an AUROC of 0.81(sensitivity: 0.86, specificity: 0.66) while the more invasive percutaneous liver biopsy histology was 0.84 (sensitivity: 0.88, specificity: 0.76). In conclusion, this study probably completes its own cycle from enthusiastic identification to realistic recognition. It is not the perfect test just yet and methodological inconsistencies have to be ironed out, but it may reduce prevarication, prompting earlier referral, and trigger actual surgical action—the only thing that makes a difference.