This study systematically reviewed the evidence on the role of urinary 5-HIAA in the diagnosis of AA. We synthesized the results of 12 prospective studies including 724 patients with a confirmed diagnosis of AA and 743 controls, and performed 2 different meta-analyses: First, a random-effects meta-analysis that showed significantly higher mean values of urinary 5-HIAA in the AA group than in the control group; second, a DTA meta-analysis that showed a low-to-moderate pooled sensitivity (68.6%) and a relatively high pooled specificity (82%).
These findings, even with a limited degree of evidence, suggest that urinary 5-HIAA could be a promising marker for the diagnosis of AA.
Regarding the diagnostic performance of 5-HIAA for the discrimination between AA and controls, the wide range of AUC, sensitivity, and specificity reported are striking. A reasonable justification for this variability is the heterogeneity in the control groups of the different studies (including healthy patients, negative appendectomies, other intraoperative surgical diagnoses, and non-surgical abdominal pain). The presence of a control group of healthy patients will always tend to overstate the diagnostic performance of the biomarker in this context, compared to a control group of patients with acute abdominal pain, in whom there may always be a certain level of systemic inflammation and metabolic stress that leads to an elevation of the biomarker. This heterogeneity in the control groups would also explain the high between-study heterogeneity observed in the meta-analysis (I2 = 97%).
We believe that the pathophysiological premise that justifies the evaluation of 5-HIAA as a potential diagnostic marker in AA is valid, as demonstrated by the random effects meta-analysis performed: urinary 5-HIAA levels are significantly higher in the AA group than in the control group in the meta-analytical model proposed. However, this does not necessarily translate into adequate diagnostic performance. Because of this, we performed a specific DTA-type meta-analysis. It should be noted that there is a potential selection bias in this analysis, given that 4 of the 12 papers in this review were not included due to a lack of data.
When we explored the potential sources of heterogeneity in this work, we found (apart from the selection of the control group) an important variability in the determination techniques employed: HPLC and ELISA are radically different techniques both in terms of precision and cost. We believe that whenever a diagnostic test is evaluated in the context of a prevalent and time-dependent pathology, such as AA, a simple, inexpensive, and rapid determination methodology should be primed, and in this case, ELISA is probably the most appropriate technique. On the other hand, it is curious that except for the work of Bosak et al. [22], which did use HPLC, since 2012 all the reported works have used ELISA. However, we have not found any chronological pattern in the diagnostic yields published for this molecule in this review.
Another aspect to consider when evaluating a diagnostic test is the limitations that the patient's baseline situation may present for its use. For example, Ca 125 has been evaluated as a potential biomarker in the context of AA, but its potential elevation due to multiple other causes (such as gynecological pathology) justifies its poor diagnostic performance [28]. In this case, it should be considered that multiple medical, pharmacological, and dietary conditions can elevate urinary 5-HIAA levels, and this is a major limitation to the use of this molecule as a routine diagnostic test in AA.
During the course of this work, we have found inconsistencies in various papers that should be explicitly mentioned to adequately interpret the results. In the case of Jangjoo et al. [23], the authors report having excluded from the analysis the outliers’ values of 5-HIAA obtained. This, in our opinion, is an important source of bias, given that many previously reported biomarkers, such as serum IL-6, may present extreme values in the context of AA and should be known for a proper understanding of the behavior of the molecule in this disease [4]. Additionally, it is pertinent to highlight that some of the studies do not adequately clarify what happens at the classification level with negative appendectomies or with patients who are initially classified as non-surgical abdominal pain and subsequently develop acute appendicitis, which constitutes an important source of confusion [25].
The main strengths of this work are its rigorous methodology, following international PRISMA guidelines and applying the QUADAS2 bias assessment tool [29, 30], and the use of a specific meta-analytical model, such as the DTA. However, this study is not exempt from limitations. 1) the sample size included is limited and the populations are heterogeneous. 2) not all the working groups have considered the potential factors that could have elevated urinary 5-HIAA 3) The absence of data has led to a potential bias when performing the meta-analyses since not all the studies have been included 4) Although in the case of urinary LRG1, multiple studies reported their results adjusted for urinary creatinine, in this review only 1 publication made this adjustment 5) a small percentage of the numerical data included in this study have been obtained using an inferential methodology and there may some inaccuracies.
In conclusion, although the evidence is very limited, urinary 5-HIAA is emerging as a potential diagnostic tool for AA. Future prospective studies with a large sample size and with an adequate assessment of factors that may elevate serotonin before the determination of urinary 5-HIAA levels should be conducted. Urinary 5-HIAA adjustment for creatinine, previously reported in the case of urinary LRG1, is an interesting aspect to evaluate in the future for this type of diagnostic biomarker. Urinary 5-HIAA does not seem to be a useful biomarker to distinguish between NCAA and CAA.