🔍 How to Read This Page — Evidence Provenance
AIN has no randomized controlled trials guiding treatment, with one exception noted below. Rather than blur that, every substantive claim on this page is tagged:
Anchored to a named, citable study with a PMID. The number came from the paper.
Standard nephrology teaching, pharmacovigilance reporting, and clinical experience. Real and useful — but not trial-anchored. Treat as a differential-generating checklist.
⚠️ Why this matters here more than usual
The entire second-line drug literature in AIN rests on case series in the single digits. When you see PRACTICE-BASED on a treatment recommendation, that is not a hedge — it is the actual state of the field.
🔍 Definition & Recognition
Acute interstitial nephritis = immune-mediated inflammation of the renal interstitium and tubules, with the glomeruli spared. It accounts for 5–15% of AKI in hospitalized patients.
The mechanism is T-cell mediated delayed-type hypersensitivity — not dose-dependent, and it recurs rapidly on re-exposure. Remember this one fact; it drives the entire treatment section, including why B-cell drugs like rituximab do not belong here.
⚠️ The latency trap
Do not anchor on “2 to 8 weeks.” Latency is agent-specific: antibiotics 1–3 weeks, PPIs 3–6 months, checkpoint inhibitors 3–12 months, mesalamine months to years. The most commonly missed AIN is a slow creatinine drift on a chronic PPI, written off as CKD progression.
💊 Causes 1: Drugs (70–75% of all AIN)
EVIDENCE-BASED — class split from pooled biopsy seriesAntibiotics
1–3 week latency
PPIs
3–6 month latency
NSAIDs
Steroid-unresponsive
Omeprazole is the single most implicated agent (approximately 12% of drug-induced AIN); amoxicillin second (approximately 8%).
PRACTICE-BASED — agent lists below💉 Antibiotics
- Beta-lactams — penicillins, all cephalosporin generations, carbapenems. Cross-reactive: avoid both classes after either
- Sulfonamides — TMP-SMX, sulfadiazine. Remember this when picking PJP prophylaxis
- Fluoroquinolones — ciprofloxacin most reported; can follow a single dose
- Vancomycin — overlapping AIN and ATN on biopsy
- Rifampin — distinct mechanism (anti-rifampin antibodies); flu-like illness, hemolysis, often dialysis-requiring
- Macrolides, tetracyclines, linezolid, daptomycin
🪨 Gastrointestinal
- All PPIs — omeprazole, esomeprazole, pantoprazole, lansoprazole, rabeprazole, dexlansoprazole. Cross-reactivity is near-universal — never switch to another PPI. Use famotidine
- 5-ASA agents — mesalamine, sulfasalazine, balsalazide. Insidious over months to years and frequently irreversible. Needs baseline and periodic creatinine
- H2 blockers — cimetidine, ranitidine (uncommon)
💦 Analgesics
- All non-selective NSAIDs and COX-2 selective agents
- Topical NSAIDs are systemically absorbed — patients and clinicians both discount these
- May carry nephrotic-range proteinuria from minimal-change overlap
🏹 Immuno-Oncology
- PD-1 — pembrolizumab, nivolumab, cemiplimab
- PD-L1 — atezolizumab, durvalumab, avelumab
- CTLA-4 — ipilimumab. Combination therapy carries the highest risk
- Frequently no rash and no eosinophilia
⚖️ Other Agents
- Allopurinol — among the most severe. Higher risk with CKD, concurrent thiazides, and HLA-B*58:01 carriers
- Thiazides and loop diuretics (sulfonamide-based)
- Aromatic anticonvulsants — phenytoin, carbamazepine, phenobarbital; cross-reactive, switch to a non-aromatic agent
- Hydralazine (also drug-induced lupus/ANCA), statins, ACEi/ARB (rare)
🔥 DRESS Triggers
Renal involvement in DRESS is AIN. The high-risk agents cluster tightly:
Allopurinol · aromatic anticonvulsants · sulfonamides · vancomycin · minocycline · dapsone · abacavir · nevirapine
Latency 2–8 weeks — longer than ordinary drug hypersensitivity. Needs longer steroid courses and relapses on fast tapers.
🎯 The checkpoint-inhibitor attribution trap
Most patients on a checkpoint inhibitor are also on a PPI. Before you attribute AIN to a cancer drug the patient needs — and stop it — account for the PPI they do not need. Stopping the PPI is free. Stopping immunotherapy is not.
🦠 Causes 2: Infection, Autoimmune, and the Rest
PRACTICE-BASED🦠 Infection (5–10%)
Bacterial: Streptococcus, Staphylococcus, Legionella, Brucella, Salmonella, Mycoplasma, Leptospira, TB, syphilis, Q fever
Viral: EBV, CMV, HIV, hantavirus, BK polyomavirus, adenovirus, parvovirus B19, hepatitis B/C, SARS-CoV-2
Fungal/parasitic: Histoplasma, Coccidioides, Leishmania, Toxoplasma
🧿 Systemic / Autoimmune
- Sarcoidosis — granulomas, hypercalcemia
- Sjögren — distal RTA, hypokalemia
- SLE — can be isolated interstitial disease
- IgG4-related disease — storiform fibrosis, mass lesions
- TINU — bilateral anterior uveitis, young women
- Anti-TBM and ABBA disease
- IBD — as an extraintestinal manifestation
🧬 Malignancy & Genetic
- Lymphoma/leukemia — direct interstitial infiltration; monoclonality distinguishes it
- Myeloma/light chain — check free light chains
- Karyomegalic interstitial nephritis — FAN1 mutations, enlarged tubular nuclei
- ADTKD — UMOD, MUC1, REN, HNF1B
⚠️ The IBD fork — get this right
A Crohn or UC patient on mesalamine with interstitial nephritis has two possible causes pointing in opposite therapeutic directions: mesalamine toxicity, where you stop the drug — or an extraintestinal manifestation of active IBD, where you intensify treatment. Disease activity, drug timing, and biopsy separate them. Guessing wrong makes the patient worse either way.
🔬 Diagnosis
The classic triad appears in under 10% of cases
Fever, maculopapular rash, peripheral eosinophilia. Absent in over two-thirds of PPI-AIN. Its absence does not exclude AIN — and waiting for it is how the diagnosis gets missed.
✅ What actually helps — the urine sediment
- Sterile pyuria + WBC casts — the most useful bedside finding
- Sub-nephrotic proteinuria, usually under 1 g/day. Nephrotic range suggests NSAID-associated minimal change
- Cast logic: RBC casts = GN · WBC casts = AIN · muddy brown casts = ATN
🚫 Urine eosinophils — a debunked test. Do not order.
EVIDENCE-BASED — PMID 24052222Muriithi et al. (CJASN 2013) tested it properly — 566 patients with both a urine eosinophil test and a native kidney biopsy within a week of each other, 91 with AIN. At a 1% Hansel-stain cutoff:
Sensitivity
Specificity
PPV
LR+
LR−
The likelihood ratios are the whole argument. Both sit at 1.0, so the result moves post-test probability essentially nowhere in either direction. A test that cannot revise your estimate is not a diagnostic test. It fails in both directions — a negative does not rule out, a positive does not rule in.
Why older sources disagree: the four largest earlier series reported sensitivity 40–91% and specificity 52–95% — but none used kidney biopsy as the gold standard.
🔭 Urinary CXCL9 — the emerging replacement
EVIDENCE-BASED — PMID 37395276Tubular epithelium secretes CXCL9 in response to IFN-gamma from infiltrating T cells — so it tracks the actual lesion. Stable at room temperature, unlike beta-2 microglobulin. Identified and validated by Moledina et al. in J Clin Invest. Availability outside research and reference labs remains limited.
🧪 What to demand from the biopsy report
Three items drive management more than the creatinine does:
- Degree of interstitial fibrosis and tubular atrophy (IFTA)
- Presence or absence of an active inflammatory infiltrate
- Granulomas — raises sarcoid, TB, fungal, PPI, fluoroquinolones
Light microscopy: dense infiltrate, tubulitis, interstitial edema, normal glomeruli and vessels. IF usually negative — except linear TBM staining in anti-TBM disease and IgG4-rich plasma cells in IgG4 disease.
💊 Treatment — Sorted by Etiology
Step 1 — Universal, every case
- Withdraw the offending agent. Sequentially if AKI is mild, simultaneously if severe
- Avoid the whole class and cross-reactive classes
- Hunt the OTC exposures — PPIs and NSAIDs go unreported because patients do not consider them medications
- Correct the tubular consequences; remove other nephrotoxins
Do not delay withdrawal awaiting biopsy. Withdrawal is reversible. The fibrosis from a two-week delay is not.
| Etiology | First line | Then | Do NOT |
|---|---|---|---|
| Drug-induced (general) | Withdraw; prednisone 1 mg/kg/day if severe or not recovering | Extend steroids to 8–12 weeks if active infiltrate → then MMF 1–2 g/day | Escalate when IFTA is heavy and infiltrate absent |
| NSAID-induced | Withdrawal + supportive only | — | Give steroids — this subtype does not respond |
| PPI-induced | Withdraw; famotidine if acid suppression needed | Steroids if not recovering → MMF | Switch to another PPI |
| Mesalamine / 5-ASA | Withdraw permanently; steroids if inflammation active | — | Rechallenge; confuse with IBD-associated nephritis |
| Allopurinol / DRESS | Withdraw; prednisone 1 mg/kg/day, longer course with slow taper | MMF or cyclosporine if steroid-dependent | Rechallenge; use a 2-week taper (DRESS relapses) |
| Checkpoint inhibitor | Hold ICI; prednisone 0.5–1 mg/kg/day | MMF → infliximab or cyclophosphamide (case-level) | Stop the ICI before excluding a concurrent PPI |
| Infection-associated | Treat the organism | Supportive | Give steroids — generally contraindicated |
| BK nephropathy (allograft) | Reduce immunosuppression | IVIG, leflunomide, cidofovir (weak evidence) | Add steroids for “rejection” without excluding BK |
| Sarcoidosis | Prednisone 0.5–1 mg/kg/day, slow taper over 6–12 months | Methotrexate / azathioprine / MMF → infliximab | Reach for rituximab; forget the hypercalcemia |
| Sjögren | Prednisone | MMF or azathioprine; HCQ adjunct | Neglect the distal RTA — bicarbonate and K⁺ repletion are treatment |
| IgG4-related disease | Prednisone — but most flare on taper | Rituximab or inebilizumab; obexelimab | Expect steroids alone to hold remission; delay (fibrotic disease will not respond) |
| TINU | Systemic steroids + topical steroids for the uveitis | MMF or azathioprine for relapsing uveitis | Manage without ophthalmology — the courses run independently |
| Anti-TBM / ABBA | Prednisone | Rituximab — mechanistically rational here (antibody-mediated) | — |
| Lymphoma / myeloma | Treat the malignancy | — | Immunosuppress for “AIN” |
| Karyomegalic / genetic | Supportive; CKD management | Transplant | Immunosuppress — no role |
⛔ The Non-Responder at 3–4 Weeks
If the patient has not recovered 3–4 weeks after withdrawal plus steroids, work it in this order:
- Question the diagnosis. Consider alternative AKI etiologies, especially if no biopsy was done. Biopsy now if feasible.
- If biopsy-confirmed, read the chronicity. This is the decision point.
- If there IS an active infiltrate — extend steroids to a total of 8–12 weeks. Do not add a second agent yet.
🛑 The most commonly missed decision in AIN
Severe chronic changes (heavy IFTA) with no acute inflammatory infiltrate: the patient will not improve. Taper and stop the glucocorticoids. Adding immunosuppression here purchases infection risk, not GFR.
💉 Second-Line Immunosuppression
Mycophenolate mofetil — the best-supported option
EVIDENCE-BASED — PMID 17699278- Dose: 1 g/day divided, titrate to 2 g/day if tolerated
- Preddie et al. (CJASN 2006) — n=8, treated 13–34 months. These patients had responded to steroids but could not tolerate withdrawal — steroid-dependent, not steroid-resistant. All discontinued glucocorticoids
- Critical limitation: only two of the eight had drug-induced AIN. Applying this to classic drug AIN is a genuine extrapolation
❌ Rituximab — where it belongs and where it does not
Not for drug-induced AIN. Drug-induced AIN is T-cell mediated, so B-cell depletion targets the wrong effector. Major reference sources do not list it for this indication.
Rational only when the driver is antibody- or B-cell-mediated: IgG4-related disease (best evidence), anti-TBM disease, ABBA disease, and at case-report level Sjögren-associated TIN.
Not standard for sarcoid — that pathway runs methotrexate, azathioprine, or MMF, then anti-TNF.
⚠️ The PJP Prophylaxis Trap
Prophylaxis applies at prednisone 20 mg/day or more for four or more weeks, especially alongside MMF. TMP-SMX is the default agent — and here it carries two specific problems:
If sulfa is anywhere on the suspect list, do not use it.
It raises creatinine 0.1–0.3 mg/dL with no change in true GFR — contaminating the exact measurement you are using to judge treatment response.
✅ Atovaquone 1500 mg daily avoids both. Dapsone requires G6PD testing first — and dapsone is itself a DRESS trigger.
🎯 Key Takeaways
🔍 Recognition
- Classic triad in under 10% — do not wait for it
- Latency is agent-specific, not a fixed window
- Sterile pyuria + WBC casts is the useful finding
- Hunt the OTC PPI and NSAID
⚡ Practice-Changing
- Do NOT order urine eosinophils — LR 0.97/1.01
- NSAID-AIN does not respond to steroids
- Never switch to a different PPI
- Rituximab has no role in ordinary drug AIN
📋 Management
- Withdrawal is the therapy — do not await biopsy
- Extend steroids to 8–12 weeks before adding a drug
- Heavy IFTA + no infiltrate = stop, do not escalate
- Atovaquone, not Bactrim, for PJP prophylaxis
📈 Prognosis
Complete recovery
Partial recovery
Progress to ESKD
Predictors of poor recovery
- Kidney failure lasting more than three weeks before treatment
- NSAID as the causative agent
- Interstitial granulomas, fibrosis, and tubular atrophy on biopsy
- Mesalamine as the agent — frequently irreversible
The one thing to take away: time-to-withdrawal is the dominant modifiable variable. Steroids may accelerate recovery when started early; nothing rescues a kidney that has already fibrosed. When deciding whether to escalate immunosuppression, the biopsy chronicity — not the creatinine — casts the deciding vote.