The quest for better and safer diagnostic tools for appendicitis and complicated appendicitis in children is surely a great task, engaging many researchers and clinicians worldwide. To date, inflammatory blood markers as well as imaging studies such as ultrasonography or are often utilized in addition to clinical assessment of a child with acute abdominal pain [20, 21]. However, the diagnostic accuracies of standard inflammatory markers vary [20], and imaging studies are not always available and may be user dependent [22]. The present study sheds some new and important light on the diagnostic utility of serum and urine LRG1. Our results indicate that urine LRG1 might have the potential to differentiate between uncomplicated and complicated appendicitis in children, but with a lesser accuracy than currently more readily available diagnostic aids such as CRP and AIR score. Additionally, urine LRG had a high specificity and PPV for diagnosing appendicitis among all children with suspected appendicitis. Serum LRG1, on the contrary, does not seem to have a role in the assessment of a pediatric patient with acute abdominal pain.

To our knowledge, only 4 previous studies have investigated the diagnostic performance of serum LRG1 for pediatric appendicitis [10, 23,24,25]. All but one [23] of these studies have control groups consisting of patients with non-appendicitis acute abdominal pain, much like in our study. Still, their results indicate a good diagnostic performance of serum LRG1, especially one in which there was no overlap between serum LRG1 concentrations in the patients with and without appendicitis, generating an astonishing ROC AUC of 1.0 [24]. However, when evaluated on a small adult population (n = 28), serum LRG1 did not differ between the ones with and without appendicitis [26]. We can only speculate on why our results are fundamentally different from previous studies on pediatric patients. First, we must consider the study population. In our cohort, an absolute majority of the patients (78%) were diagnosed with appendicitis – a much greater proportion than in the previous studies [12]. This indicates that the threshold for referral of patients from either a pediatrician or a nurse at the pediatric ED, or from a family practician, to the on-call pediatric surgeon due to suspected appendicitis was quite high. And hence, even among the patients in the non-appendicitis group, the suspicion of appendicitis was high, even though most of them were diagnosed with unspecified abdominal pain. Worthy of mentioning is that the digital medical record system at our hospital includes all hospitals in the region, and that we therefor are certain that none of the patients sent home were diagnosed with appendicitis during the weeks after inclusion in the study, at least not in the region. This does not, however, exclude the possibility of spontaneously resolving appendicitis among these patients, who in that case would have been misclassified.

Another possible explanation for the vastly divergent results of different studies on LRG1 and appendicitis is the large selection of commercially available test kits used [12]. To our knowledge, the ELISA test kit used in this study (Cusabio, Hubei Province, China, CSB-E12962 h) has only been used in one previous appendicitis study – namely one from our group [14]. This previous study showed promising diagnostic properties of urine LRG1, especially when combined with urine creatinine. Although the study population of our previous study was substantially smaller than the current one, this indicates that the use of different test kits alone cannot entirely explain the different results.

In our cohort, urine LRG1 had a high specificity and PPV for diagnosing appendicitis among all children with suspected appendicitis. Unfortunately, the sensitivity and NPV in this setting were disappointingly low. Urine LRG1s ability to correctly identify the patients with complicated disease was better than both leukocytes, neutrophils and percentage neutrophils, but poorer than CRP and AIR score. Still, even though an AUC over 0.5 is indicative of some discriminative capacity, generally only values above 0.7 are considered acceptable [27]. The rationale behind our three AUC values below 0.5, is that the median concentrations of serum and urine LRG1 in the appendicitis were lower than in the non-appendicitis, and that the median concentration of serum LRG1 was lower in the complicated appendicitis coup compared to the uncomplicated appendicitis group, which is also reflected in the negative associations from the logistic regression analyses. Based on the results from previous studies indicating a positive association between increased LRG1 concentrations and appendicitis, the test was set up in a similar matter and not inversed.

Only a few previous studies have evaluated both the diagnostic and discriminative performances of urine LRG1. Kakar et al. concluded that the diagnostic performance of urine LRG1 (AUC 0.7) were better than its discriminative performance (AUC 0.6) [23]. In this study, however, the non-appendicitis group was comprised of patients who did not have acute abdominal pain, but instead different traumatic and non-inflammatory conditions – a setting rather far from the clinical one. Kharbanda et al. found that urine LRG1 in was significantly higher among pediatric patients with perforated appendicitis compared to those with non-perforated appendicitis. Unfortunately, they do not seem to have investigated the discriminative performance of LRG1 in this setting further. They did however conclude that leukocytes performed better than serum and urine LRG1 in diagnosing appendicitis [10].

As previously mentioned, only a small proportion of the included patients had non-appendicitis abdominal pain, and the predictive values of the different biomarkers need to be interpreted in the light of the high appendicitis prevalence. One must also consider if our current study population accurately reflects the intended target population for LRG1 utility. The results of our study do not support an introduction of LRG1 in clinical practice to help differentiate between uncomplicated and complicated appendicitis, since both CRP and AIR scored showed better discriminative performances. Furthermore, one could argue the great advantage of urine LRG1 being non-invasive is somewhat lost if the patient must go through venipuncture anyhow, for example to obtain preoperative blood tests. And in the hospital setting, further evaluations such as ultrasonography are often carried out. Apart from being diagnostic, ultrasonography can also identify appendicoliths, which are associated with increased risk of complicated appendicitis [28]. It is therefore possible that urine LRG1 could be best used to rule out appendicitis in patients with a low pretest probability, and evaluation of the biomarker in this setting, for example at a primary care facility, would be of great interest.

The main strength of the study lies in the prospective study design, with standardized collection of data and clear definitions of appendicitis and its’ different severities. Another strength is that the cohort is relatively substantial compared to previous studies on LRG1. Of course, the results of the current study must be interpreted in the light of its limitations. The first one includes the quite high number (n = 43) of excluded patients, for which we don’t have demographic data. This introduces the potential of selection bias. Secondly, we did not analyze urine creatinine and were therefore unfortunately not able to adjust our results for possible dehydration – something that has been shown to increase the diagnostic performance of urine LRG1 in previous studies [14].