Breath Ammonia Moves to the Bedside in Diabetic Kidney Disease Screening
A new observational study from Shanghai Pudong Hospital is testing whether exhaled ammonia can serve as a noninvasive marker of diabetic kidney disease in adults with type 2 diabetes. The premise is straightforward but clinically ambitious: if renal dysfunction alters ammonia handling in a predictable way, a single breath sample could help flag kidney injury before routine care misses the window for action.
The study is built around a wearable NH₃ sensor designed for remote monitoring and diagnosis. Investigators plan to pair breath measurements with standard kidney assessments, including serum creatinine, estimated glomerular filtration rate, and urinary albumin-to-creatinine ratio, then use those data to construct and validate a diagnostic model. In practical terms, the study asks whether a volatile metabolic signal in expired air can track with kidney disease stage closely enough to influence screening workflows in diabetes clinics.
A total of 380 participants have been enrolled so far, although the protocol describes a broader target population of roughly 1,500 to 2,000 adults with type 2 diabetes. Eligibility is limited to adults 18 years and older who can cooperate with testing and provide informed consent. Participants with mental disorders preventing cooperation and those with known allergy or contraindication to the sensor materials or sampling process are excluded.
How the Breath Test Is Being Built
The protocol uses a hybrid design: a cross-sectional observational component for model construction and a prospective cohort component for external validation. Participants are stratified by KDIGO 2023 criteria into a diabetic kidney disease group and a non-DKD diabetes group. DKD is defined here by UACR ≥30 mg/g and eGFR 25-75 mL/min/1.73 m², which places the focus on clinically relevant renal impairment rather than subtle laboratory drift.
After an 8-hour fast, participants collect end-tidal breath in disposable sterile tubes. The sensor then measures ammonia concentration in parts per billion. At the same time, fasting glucose, HbA1c, lipid profile, serum creatinine, eGFR, and UACR are recorded. This pairing matters because it allows breath ammonia to be judged against established renal and metabolic markers instead of being treated as an isolated signal.
The study’s diagnostic plan is statistical rather than speculative. Logistic regression will be used to build models based on NH₃ alone and then in combination with UACR and eGFR. Receiver operating characteristic curves will generate AUC, sensitivity, and specificity, and the optimal threshold will be selected using the Youden index. Model stability will be tested with 10-fold cross-validation and an external validation cohort.
A single breath sample is being asked to do the work of several blood and urine tests.
Why Ammonia Could Track Renal Injury
The biological rationale is plausible. The kidney is central to nitrogen handling, acid-base balance, and the clearance of metabolites that can influence circulating and exhaled ammonia. As nephron function declines, the metabolic footprint of kidney disease changes, and those shifts may be detectable outside the bloodstream. If ammonia concentration in exhaled breath correlates inversely with eGFR and positively with albuminuria, it could offer a physiologic readout of renal stress that is easier to repeat than a blood draw.
That logic is especially relevant in diabetes, where kidney injury often progresses silently. Patients may carry a substantial burden of renal damage before symptoms appear, and conventional monitoring can miss opportunities for earlier risk stratification. A breath-based sensor would not replace creatinine, eGFR, or UACR, but it could function as a rapid screening layer in clinics where repeated laboratory testing is burdensome or delayed.
The protocol also anticipates a three-tier renal health risk system based on exhaled ammonia concentration. Low-, intermediate-, and high-risk categories would be linked to management pathways, with the stated goal of improving screening coverage and early diagnosis. In a real-world diabetes program, that kind of triage could influence who gets prioritized for confirmatory kidney testing, nephrology referral, and intensified metabolic control.
Device Performance and the Gap Between Signal and Usability
Beyond diagnostic accuracy, the study defines technical requirements for the wearable sensor itself. The device is expected to achieve a detection limit of ≤0.5 ppb, a response time of ≤10 seconds, humidity and temperature interference below 5%, and power consumption of ≤10 mW. It is also described as needing to meet ISO 13485 requirements, which points to an intent to align the sensor with medical device quality standards rather than consumer-grade electronics.
Those engineering targets are not decorative. Breath analysis is vulnerable to environmental noise, and a useful point-of-care device has to perform reliably in the messy conditions of routine clinical care. Humidity, ambient temperature, and sampling variability can all distort volatile compound detection, so the study’s technical metrics matter as much as the clinical cutoff values. If the sensor cannot remain stable across common clinical settings, even a strong association with kidney disease would be hard to translate into practice.
The protocol also includes a health economic evaluation, with projected reductions in end-stage renal disease incidence and annual medical cost savings. A market penetration target is listed as well. Those measures will not determine whether the sensor works biologically, but they signal a broader intent: the investigators are trying to position breath ammonia monitoring as a screening tool that could scale in routine care if validation succeeds.
What the Record Shows Now, and What It Does Not
At present, the record documents an ongoing observational study rather than clinical outcomes. No results are posted, so the key question remains whether exhaled ammonia will separate DKD from non-DKD diabetes with the accuracy thresholds built into the protocol. The record sets a high bar for performance, including AUC ≥0.92, sensitivity of at least 90%, and specificity of at least 85%.
Those thresholds are ambitious, and they will need to be interpreted against the realities of diabetic kidney disease, where renal impairment exists on a continuum and breath biomarkers may be influenced by metabolic status, fasting state, and comorbid illness. The planned external validation cohort will be especially important because a model that performs well in one hospital population can lose reliability once it is tested elsewhere.
Data sharing is planned 12 months after the primary results publication, with protocol, statistical analysis plan, informed consent form, clinical study report, and analytic code listed for sharing. Until the study reports its findings, the clinical value of wearable breath ammonia monitoring remains a testable hypothesis rather than a completed diagnostic advance.