Back
Acute interstitial nephritis (AIN) is an important cause of acute kidney injury (AKI) that has been historically associated with the presence of urine eosinophils (UE). In more recent studies, this association between AIN and UE has grown weaker and, to our knowledge, only one study used kidney biopsy as the diagnosis standard. In recent years, new AKI and AIN urine biomarkers, such as interleukin-9 and tumour necrosis factor-α, have gained interest, yet these are not widely available. To investigate the performance of UE screening test in cases of suspected AIN, we evaluated their association with biopsy proven AIN and whether UE was associated with the decision to biopsy.
We conducted a retrospective study with adult patients who underwent both UE screening and native kidney biopsy performed in a maximum seven days period at Centro Hospitalar Universitário de Santo António between 2010 and 2020. We also analysed the patients with UE screening and no biopsy. We considered UE positive screening when >1% of eosinophils were seen in citospin smears of urinary sediment, using Leishman staining. Samples with less than two white blood cells (WBC) per high power field in urinary sediment (< 10 WBC/µL) were considered negative. Data was analysed using Microsoft Excel©, chi-square test was used to determine association and a p-value of 0.001 was considered statistically significant.
We have identified 1686 requests for UE screening, after exclusion of duplicated orders and patients that didn’t meet inclusion criteria. From the 1686 patients, 105 patients (6,2 %) underwent biopsy: 47 females and 58 males, with ages between 18 and 88 years old and median age of 63 years old. UE were present in 18 urinary sediments and AIN in 15 biopsies. Among the patients with UE positive screening, three had AIN, seven proliferative glomerular disease, three non-inflammatory glomerular disease, one had both non-inflammatory glomerular disease and diabetic kidney disease, one had chronicity findings, one had hypertensive arterionephrosclerosis and one had crescentic glomerulonephritis. The sensitivity of UE for AIN was 20,00% and specificity was 83,33%. The positive predictive value was 16,67% with a negative predictive value was 86,21%.
From the 1686 patients, 180 had >1% UE. Using chi-square test, we obtained a p value of 0,027, determining that there was no association between UE presence and the decision to biopsy.
Urine eosinophils were found in a wide range of biopsy proven kidney diseases, and they demonstrated a low sensibility and positive predictive value for AIN. These findings are in line with the only robust data published and with the experts’ opinion. Although they exhibited higher specificity and negative predictive values, they do not allow the diagnosis or exclusion of AIN. Since the presence of urinary eosinophils does not alter the decision to carry out a kidney biopsy and it is not a sensitive or specific test, the diagnosis and therapeutic decisions should be supported by kidney biopsy.