Dear Editor,

We read the article published by Makay et al. [1] with a great interest. They suggested that high neutrophil-to-lymphocyte ratio (NLR) might be a marker to predict gastrointestinal (GI) bleeding in Henoch–Schönlein purpura (HSP). As known, acute hemorrhage is associated with leukocytosis and neutrophilia. During the first 1–3 h of an acute hemorrhage, neutrophilia occurs because of a shift from the marginal pool to the circulating pool [2]. Considering this suggestion, we evaluated the predictive role of NLR for organ involvement in HSP patients without GI bleeding.

We reviewed the medical files of 217 previously healthy patients who were diagnosed as HSP with EULAR/PRES criteria and hadn’t been treated yet. Demographic data, symptoms, signs of disease and laboratory data at the time of diagnosis were recorded from patients’ file and computerized database. Patients with active GI bleeding were excluded, and the rest were divided into two groups. While group 1 included GI and/or renal involvement and arthritis accompanying palpable purpura, group 2 included patients with only arthritis and palpable purpura. Then, NLR was evaluated in order to determine whether it’s a suitable marker for predicting organ involvement. Variables were expressed as medians and 25th and 75th percentile. Differences between the NLR values according to the groups were assessed by Mann–Whitney U test. Receiver operating characteristic (ROC) curves were used to find out a cutoff value for predicting GI and/or renal involvement in HSP.

The numbers and NLR median values with respect to the involvements of patients are shown in Fig. 1, and demographic and laboratory findings are summarized in Table 1. While no relationship between sex and organ involvement was detected, the age of group 1 was found significantly high. As shown in Table 1, statistically significant level of higher NLR was recorded in group 1 than in group 2. The area under the ROC curve (AUCROC) analysis showed that NLR could be a fair predictor of GI and/or renal involvement in HSP (AUCROC = 0.731, 95 % CI 0.66–0.79, p = 0.000). A cutoff NLR value was found as 3.09 for GI and/or renal involvement with 61 % sensitivity and 78 % specificity.

Fig. 1
figure 1

The numbers of patients and neutrophil-to-lymphocyte ratio (median) values with respect to involvements

Full size image
Table 1 Demographic and laboratory findings
Full size table

With this retrospective study, it was revealed that high NLR level is related with GI and/or renal involvement in HSP. However, the role of NLR to foresee only renal involvement couldn’t be statistically evaluated due to very small number of patients (n = 2) who have only renal involvement. Makay et al. reported that high NLR might predict GI bleeding in HSP, and they suggested 2.82 as an optimal cutoff NLR value for predicting GI bleeding with 81 % sensitivity and 76 % specificity. But according to our hypothesis, establishing a relationship between NLR and disease severity is not a valid approach in HSP patients with GI bleeding since the cause of the increase in NLR level cannot be differentiated. It could be either raised from the inflammation or GI bleeding. Although the cutoff value obtained from our study group was not enough sensitive and specific to predict organ involvement, as the patients with GI bleeding were excluded, the results were assessed valuable. However, our study is limited by having a retrospective design and a small number of renal involvements.

In conclusion, NLR is a weak predictor of organ involvement in HSP; in this respect, the clinical findings and clinician’s attention are more precious as always.