Positive predictive value of urine eosinophils for AIN

AKImedium

A 52-year-old man is evaluated for AKI. He has taken omeprazole 20 mg twice daily for the last 3 weeks for indigestion. He is otherwise asymptomatic and has not observed any changes in the volume or character of his urine. BP is 138/82 mm Hg and heart rate is 86/min. Physical examination, including heart, lungs, and abdomen, is unremarkable. No edema or rash is observed. Laboratory data reveal serum creatinine 2.9 mg/dL compared with 0.9 mg/dL from 2 months ago (reference range, 0.7–1.3). Urinalysis is positive for leukocyte esterase but negative for nitrites. Urine microscopy reveals 20 WBC/HPF, and eosinophils account for >6% of the total urinary WBCs, which is considered a positive result at the reference laboratory. A urine culture shows no growth after 72 hours. Despite withdrawal of omeprazole, 5 days later, the creatinine remains 3.0 mg/dL. Acute interstitial nephritis (AIN) is suspected, and a kidney biopsy is recommended. The patient asks if a biopsy is necessary given the presence of urinary eosinophils. To address this question, the physician identifies a study of patients with pyuria and AKI, sufficiently similar to this patient, who had both a urine eosinophil test and a kidney biopsy performed within the same week. A table of results is constructed for the 467 patients with pyuria: urine eosinophils >5%, 18 with biopsy showing AIN and 42 with another biopsy result (total 60); urine eosinophils ≤5%, 55 with AIN and 352 with another result (total 407); totals, 73 with AIN and 394 with another result (467 overall). Based on this study, what is the probability that a kidney biopsy performed in this patient will demonstrate AIN?

  1. A.30%Correct
  2. B.16%
  3. C.25%
  4. D.86%

Explanation

The probability that a kidney biopsy in this patient will show AIN, based on these data, is 30%. AIN is an important cause of AKI, for which eosinophiluria has been proposed as a noninvasive diagnostic strategy. Nephrologists must have a sufficient understanding of the principles of diagnostic testing to understand the limitations of this test. In this study of patients similar to the case described, the positive predictive value (PPV) of urine eosinophil count >5% is 30%. A PPV is defined as the probability that a patient with a positive test will have the disease in question. Thus, despite a clinical suspicion for AIN, 70% of patients will have an alternative diagnosis, such as glomerulonephritis, acute tubular necrosis, atheroembolic disease, or paraprotein deposition, among other possibilities. Because corticosteroids are not the best treatment for many of these conditions, kidney biopsy is often warranted. In a 2 × 2 table, a positive test includes A patients with disease and B without disease, and a negative test includes C patients with disease and D without disease; PPV = A/(A + B), NPV = D/(D + C), sensitivity = A/(A + C), and specificity = D/(D + B). As demonstrated in this representative 2 × 2 table, sensitivity and specificity are descriptors of diagnostic test performance that are independent of the prevalence of the condition in the population being assessed. Sensitivity is defined as the proportion of patients who have the condition of interest who will have a positive test result. A negative result for a highly sensitive test is useful to exclude a diagnosis. Urine eosinophil count >5% occurred in only 25% of patients found to have AIN; thus, the absence of eosinophiluria is insufficient to exclude the diagnosis. Specificity is defined as the proportion of patients without the condition of interest who will have a negative result. A test with a very high specificity can help establish a diagnosis. This study found the specificity of urine eosinophilia >5% was 90%, which supports a diagnosis of AIN but does not establish the diagnosis. Calculations of the PPV and negative predictive value (NPV) reveal the performance of a diagnostic test in the relevant population by taking prevalence into account. As described, the PPV is defined as the probability that subjects who have a positive test have the condition in question, referred to as true positives. It is influenced by the prevalence of a disease in a population. A positive test result for a highly prevalent condition will have a high PPV. NPV describes the probability that a subject with a negative test does not have the disease, referred to as true negatives. A negative test result in a population with a low prevalence of the condition in question has a high NPV. In this study in which the overall prevalence of AIN was 16%, the PPV and NPV of urine eosinophil count >5% are 30% and 86%, respectively. Neither predictive value would be adequate to rule in or out the diagnosis of AIN.

Why each option

A.
Correct. The positive predictive value is 18 of 60 patients with eosinophils above 5%, or 30%, the post-test probability of AIN.
B.
Sixteen percent is the overall prevalence of AIN (73 of 467), the pretest probability before the positive eosinophil result is considered.
C.
Twenty-five percent is the sensitivity (18 of 73 AIN cases had eosinophils above 5%), not the probability of disease given a positive test.
D.
Eighty-six percent is the negative predictive value (352 of 407), the probability of no AIN when eosinophils are 5% or less.

Reference: Muriithi AK, Clin J Am Soc Nephrol 2013 (utility of urine eosinophils in the diagnosis of acute interstitial nephritis); Perazella MA, Nat Rev Nephrol 2010 (diagnosis and treatment of AIN)

This is one of 1,100 questions in the MedBoardSA nephrology bank, with timed test mode, tutor mode, and progress tracking by topic.

Practise the full bank

More AKI questions

Browse other topics