What it is:This appendix is a drug lookup table, not a disease chapter. It catalogs which drugs have been implicated in six distinct drug-induced blood disorders: aplastic anemia, agranulocytosis, immune hemolytic anemia, oxidative (G6PD-related) hemolytic anemia, megaloblastic anemia, and thrombocytopenia.
The core problem:Drug-induced cytopenias mimic primary hematologic disease. If "new drug" never makes it onto your differential, you chase the wrong workup, order the wrong tests, and leave the patient on the drug that's causing the problem.
What you do about it:Any unexplained drop in a blood cell line gets a medication timeline before anything else. Stop the suspect drug, match your management to which cytopenia you're looking at, and document the reaction so nobody re-challenges the patient later.
Sort these six disorders into two buckets and the whole appendix stops being a memorization dump. Decreased production(the marrow is making less): aplastic anemia, agranulocytosis, megaloblastic anemia. Increased destruction(the marrow is fine, cells are getting killed off in circulation): immune hemolytic anemia, oxidative hemolytic anemia, thrombocytopenia (most drug-induced cases). Production problems tend to build slowly and show a low reticulocyte count. Destruction problems can hit fast and show a reticulocyte count that's climbing, or a positive Coombs test.
Every drug name in this appendix got there through one of a handful of mechanisms. Learn the mechanism and you can predict the timeline, the workup, and whether re-challenge is ever an option, without memorizing 300 individual drug names.
| Mechanism | What's actually happening | Timeline / predictability | Classic example |
|---|---|---|---|
| Direct dose-dependent toxicity | Drug or metabolite directly damages marrow progenitors or interferes with DNA synthesis, proportional to dose and duration | Predictable, dose-related | High-dose chemotherapy, high-dose chloramphenicol |
| Idiosyncratic marrow injury | Unpredictable individual susceptibility, not clearly dose-related | Rare, can appear at any point in therapy | Carbamazepine-induced aplastic anemia, clozapine-induced agranulocytosis |
| Hapten mechanism | Drug covalently binds a protein on the cell surface; the immune system raises an antibody against the drug-coated cell, which then gets destroyed | Needs sustained, high-dose drug exposure first; resolves once the drug clears | High-dose IV penicillin-induced hemolytic anemia |
| Immune complex ("innocent bystander") | Drug binds a circulating antibody, and the complex adsorbs onto a cell surface and fixes complement; the cell is destroyed incidentally | Can happen with tiny doses; re-exposure triggers a reaction within hours | Quinine- and quinidine-induced thrombocytopenia |
| Drug-induced autoantibody | Drug alters the cell membrane or immune regulation so the body starts making true autoantibodies against the cell itself | Slow onset (months of exposure); can persist after the drug is stopped | Methyldopa-induced hemolytic anemia |
| Oxidative injury | Drug or metabolite generates oxidative stress the cell can't buffer, especially when antioxidant capacity (G6PD/glutathione) is already low | Predictable in G6PD-deficient patients, dose and exposure dependent | Dapsone, primaquine, sulfonamides |
| Antifolate / antimetabolite interference | Drug blocks folate metabolism or DNA synthesis, so the nucleus can't mature while the cytoplasm keeps developing | Predictable, dose and duration related | Methotrexate, hydroxyurea, trimethoprim |
Hapten vs. immune complex mechanisms are the classic near-miss pair.Hapten reactions need weeks of a high, sustained dose of the same drug before antibodies build up. Immune complex reactions can fire off a tiny dose, and re-challenge produces a reaction within hours because the antibody is already circulating and waiting. If a patient reacts to a trace dose or reacts almost immediately on re-exposure, think immune complex, not hapten.
Dose-dependent, predictable reactions (chemo agents, high-dose chloramphenicol) you can often anticipate and monitor for. Idiosyncratic reactions (carbamazepine's aplastic anemia risk, clozapine's agranulocytosis risk) can't be predicted by dose or duration, which is exactly why drugs like clozapine carry mandatory REMS-level ANC monitoring instead of a "watch for symptoms" approach.
The symptoms just follow from which cell line is dropping. Know the pattern for each and you'll recognize a drug-induced cytopenia before the labs confirm it.
| Disorder | What you see |
|---|---|
| Aplastic anemia | All three lines drop together: fatigue and pallor (anemia), fever and recurrent infections (neutropenia), easy bruising, petechiae, and bleeding (thrombocytopenia). Seeing all three at once is the tell. |
| Agranulocytosis | Fever, sore throat that won't resolve, mouth ulcers, and sudden severe infection. A "sore throat that isn't getting better" on a known offending drug is a classic presenting complaint. |
| Immune / oxidative hemolytic anemia | Jaundice, dark or tea-colored urine, fatigue out of proportion to the anemia's severity, back or abdominal pain (more common with intravascular/oxidative hemolysis), and sometimes splenomegaly. |
| Megaloblastic anemia | Gradual fatigue and pallor, glossitis. Onset is slow (weeks to months of exposure), so it often reads as "this has always been there" rather than an acute problem. |
| Thrombocytopenia | Petechiae, purpura, easy bruising, mucosal bleeding (gum bleeding, epistaxis). Exception: heparin-induced thrombocytopenia (HIT) presents with new thrombosis, not bleeding, despite the falling platelet count. |
It doesn't matter how well the patient otherwise looks. A recent start of clozapine, methimazole, propylthiouracil, dapsone, or sulfasalazine plus a new fever means you treat it as neutropenic sepsis until proven otherwise, you don't wait on the ANC to come back before acting.
Always ask about any new drug started in the last 1 to 12 weeksfor suspected immune mechanisms, and any drug the patient has been on for months to yearsfor idiosyncratic marrow injury or autoantibody mechanisms. A reaction within hours of a dose, in someone previously exposed to that drug, points to an immune complex mechanism, not a hapten one.
A CBC with differential is step one for every suspected drug-induced cytopenia. What you check next depends on which line is affected.
| Disorder | Workup |
|---|---|
| Aplastic anemia | Pancytopenia with a low reticulocyte count(a production problem, not a destruction problem). Bone marrow biopsy showing a hypocellular marrow confirms it. Severe disease is generally characterized by marrow cellularity under 25% plus at least two of: ANC <500 cells/mm³, platelets <20,000/mm³, or corrected reticulocyte count <1%. |
| Agranulocytosis | Repeat CBC with differential to confirm and trend the ANC. Severe neutropenia is generally ANC <500 cells/mm³, with agranulocytosis describing the more extreme end of that range, often quoted around ANC <100 cells/mm³. |
| Immune hemolytic anemia | Positive direct antiglobulin test (DAT/Coombs)supports an immune mechanism. Supporting labs: low haptoglobin, high LDH, high indirect bilirubin. Peripheral smear: schistocytes suggest intravascular hemolysis, spherocytes suggest extravascular immune destruction. |
| Oxidative hemolytic anemia | Same hemolysis labs as above, plus peripheral smear findings of Heinz bodies and bite cells. Confirm with a G6PD enzyme assay, ideally drawn several weeks after the acute event, since a reticulocytosis during active hemolysis can falsely normalize the level. |
| Megaloblastic anemia | MCV frequently >100 fL with hypersegmented neutrophils on smear. Rule out true B12 or folate deficiency before attributing it purely to the drug, and consider that some drugs cause a secondary folate deficiency rather than directly blocking synthesis. |
| Thrombocytopenia | Platelet count <150,000/mm³. Rule out pseudothrombocytopenia (EDTA-induced platelet clumping) with a repeat count in a citrate tube before you commit to a diagnosis. For suspected HIT specifically, use a pretest probability tool before ordering antibody assays, don't reflexively send the lab on every falling platelet count on heparin. |
What "evidence tier" actually means:every table in this appendix sorts its drugs into three buckets, and that sorting maps onto formal causality assessment. Observational study evidencemeans the association shows up in cohort or registry-level data, the strongest tier here. Case report evidence (probable or definite causality rating)means individual case reports were formally scored, often with a tool like the Naranjo Adverse Drug Reaction Probability Scale, which grades a reaction as definite, probable, possible, or doubtful based on things like the temporal relationship, what happened on dechallenge and rechallenge, and whether other causes were ruled out. MedWatch postmarketing reportsare raw spontaneous reports to the FDA (2009-2020 window here), the weakest tier, since nothing has been independently verified.
A drug that only appears in the MedWatch tier isn't necessarily rarer or safer than one with observational study evidence, it may just be a newer drug that hasn't accumulated the volume of data needed to move up a tier yet. Don't read "MedWatch only" as "low risk."
The marrow stops producing all three cell lines, so red cells, white cells, and platelets all fall together (pancytopenia). Some cases are dose-dependent and predictable; the classic teaching example, chloramphenicol, actually causes two separate things: a predictable, dose-related marrow suppression during therapy, and a completely separate, non-dose-dependent idiosyncratic aplastic anemia that can appear even after the drug is stopped.
| Evidence tier | Drugs |
|---|---|
| Observational study evidence | Carbamazepine, furosemide, gold salts, mebendazole, methimazole, NSAIDs, oxyphenbutazone, penicillamine, phenobarbital, phenothiazines, phenytoin, propylthiouracil, sulfonamides, thiazides |
| Case report (probable/definite) | Acetazolamide, aspirin, captopril, chloramphenicol, chloroquine, chlorothiazide, chlorpromazine, dapsone, felbamate, interferon alfa, lisinopril, lithium, nizatidine, pentoxifylline, quinidine, sulindac, ticlopidine |
| MedWatch postmarketing | Adalimumab, aliskiren, amlodipine, carvedilol, dantrolene, etanercept, infliximab, oxcarbazepine, pembrolizumab, posaconazole, valsartan |
Management:stop the offending drug immediately. Supportive care includes transfusions for symptomatic cytopenias and prompt treatment of infection. Severe cases get referred for hematology-directed immunosuppressive therapy or hematopoietic stem cell transplant, decisions that live well outside this appendix's scope and belong to a specialist.
Don't collapse chloramphenicol's marrow effects into one thing. The dose-dependent suppression is common, reversible, and predictable. The idiosyncratic aplastic anemia is rare, often fatal, and has no relationship to dose, this is exactly the distractor pair a question writer loves.
NSAIDs, phenobarbital, phenytoin, and carbamazepine show up in nearly every table in this appendix, not just this one. If you only memorize four drugs to associate with "watch the blood counts," these are the four.
An idiosyncratic, immune-mediated destruction of neutrophil precursors or mature neutrophils. Clozapineis the drug every pharmacy student should know cold here: its agranulocytosis risk is documented well enough that it drives mandatory REMS-level ANC monitoring for every patient on the drug, not just a "watch for symptoms" warning label.
A note on this table: the source appendix prints its evidence-tier columns for this specific disorder as three side-by-side lists that didn't extract cleanly in a way that reliably preserves which drug belongs to which tier. Rather than guess and risk mis-stating how strong the evidence is for any one drug, the list below is consolidated and organized by drug class instead. Every other table in this appendix, and every other section of this document, keeps its original tier structure.
| Drug class | Reported to cause agranulocytosis |
|---|---|
| Antipsychotics | |
| Aripiprazole, chlorpromazine, clozapine, haloperidol, molindone, olanzapine, paliperidone, phenothiazines, pimozide, quetiapine, risperidone, thiothixene, ziprasidone | |
| Antithyroid agents | |
| Carbimazole, methimazole, propylthiouracil | |
| Antibiotics / antimicrobials | |
| β-lactams (ampicillin, carbenicillin, cefepime, cefotaxime, cefuroxime, imipenem-cilastatin, nafcillin, oxacillin, penicillin G, ticarcillin), macrolide antibiotics, clindamycin, chloramphenicol, gentamicin, streptomycin, rifampin, rifabutin, sulfonamides, sulfasalazine, vancomycin, linezolid, terbinafine, griseofulvin, flucytosine | |
| Anticonvulsants / CNS | |
| Carbamazepine, ethosuximide, lacosamide, lamotrigine, levetiracetam, oxcarbazepine, phenobarbital, phenytoin, primidone, valproic acid, memantine | |
| Cardiovascular | |
| Amlodipine, captopril, digoxin, dipyridamole, ethacrynic acid, hydralazine, hydrochlorothiazide, procainamide, propafenone, propranolol, quinidine, quinine, spironolactone, ticlopidine, tocainide, trandolapril | |
| Antidepressants | |
| Clomipramine, desipramine, doxepin, fluoxetine, imipramine, mirtazapine | |
| Other | |
| Acetaminophen, acetazolamide, boceprevir, colchicine, dapsone, deferasirox, ganciclovir, valganciclovir, glyburide, chlorpropamide, tolbutamide, metformin, ibrutinib, indomethacin, leflunomide, levodopa, meprobamate, methazolamide, methyldopa, metronidazole, octreotide, oseltamivir, pantoprazole, pentazocine, prednisone, pyrimethamine, ustekinumab, zidovudine, chlorpheniramine | |
Management:stop the drug. Neutropenic fever gets treated as an emergency (empiric broad-spectrum antibiotics, cultures first if it doesn't delay treatment). Granulocyte colony-stimulating factor (G-CSF) is used in select cases to speed count recovery, per the treating team's judgment.
Most idiosyncratic drug reactions rely on the clinician noticing symptoms and acting. Clozapine flips that: the ANC gets checked on a fixed schedule beforesymptoms are expected, precisely because agranulocytosis here is common enough and dangerous enough that waiting for a sore throat isn't an acceptable safety net.
Antibodies (drug-dependent or true autoantibodies) mark red cells for destruction, either in the vessel (intravascular, more dramatic) or by the spleen (extravascular, more common). This is where the hapten, immune complex, and autoantibody mechanisms from the mechanisms table show up clinically.
| Evidence tier | Drugs |
|---|---|
| Observational study evidence | Phenobarbital, phenytoin, ribavirin |
| Case report (probable/definite) | Acetaminophen, ACE inhibitors, β-lactam antibiotics, cephalosporins, ciprofloxacin, clavulanate, dabigatran, dimethyl fumarate, efavirenz, erythromycin, etoricoxib, hydrochlorothiazide, indinavir, interferon alfa, iomeprol, ketoconazole, lansoprazole, levodopa, levofloxacin, methyldopa, minocycline, NSAIDs, omeprazole, p-aminosalicylic acid, phenazopyridine, probenecid, procainamide, quinidine, rifabutin, rifampin, streptomycin, sulbactam, sulfonamides, sulfonylureas, tacrolimus, tazobactam, teicoplanin, tolbutamide, tolmetin, triamterene |
| MedWatch postmarketing | Amlodipine, bevacizumab, chlorpropamide, deferasirox, fludarabine, pegademase, pioglitazone, rosiglitazone |
Management:stop the drug. For antibody-mediated hemolysis that's actively symptomatic, corticosteroids are the mainstay; transfuse for symptomatic anemia. Severity and duration of hemolysis after stopping the drug depend heavily on which mechanism produced it.
Methyldopa doesn't need to be present anymore to keep causing hemolysis, because it induces a true autoantibody against the red cell membrane itself. That's why methyldopa-induced hemolytic anemia can be slow to resolve and why the DAT can stay positive well after the drug is stopped, unlike a hapten-mechanism reaction that clears once the drug does.
A completely different mechanism from the immune hemolytic anemia above, so don't lump them together. Here the drug or its metabolite generates oxidative stress that overwhelms the red cell's antioxidant defenses, mainly reduced glutathione, which depends on the enzyme glucose-6-phosphate dehydrogenase (G6PD). Patients with G6PD deficiency, an X-linked condition more common in people of African, Mediterranean, Middle Eastern, and Southeast Asian descent, are at the highest risk, but the mechanism can occur outside G6PD deficiency too at high enough oxidative load.
| Evidence tier | Drugs |
|---|---|
| Observational study evidence | Dapsone, rasburicase |
| Case report (probable/definite) | Ascorbic acid, metformin, methylene blue, nalidixic acid, nitrofurantoin, phenazopyridine, primaquine, sulfacetamide, sulfamethoxazole, sulfanilamide |
Management:stop the offending drug, support with transfusion if hemolysis is severe, and screen for G6PD deficiency once the acute event has resolved (testing during active hemolysis can be falsely reassuring).
Active hemolysis destroys the oldest, most enzyme-deficient red cells first, leaving behind a younger reticulocyte population with relatively normal-looking G6PD activity. Testing during the acute event can produce a falsely normal result. Wait a few weeks, or test the patient's family members if an urgent answer is needed.
Dapsone, primaquine, sulfamethoxazole, and nitrofurantoinare the four drugs most likely to show up on an exam question stem paired with "G6PD deficiency." Rasburicase belongs on this list too and is easy to forget because it's an oncology supportive-care drug, not an antibiotic.
The nucleus can't finish maturing while the cytoplasm keeps developing, because the drug is interfering with folate metabolism or DNA synthesis directly. The result is large, immature-looking red cells (MCV climbs) even though the marrow is actively trying to produce them, so this doesn't fit neatly into the "production vs. destruction" framework from the snapshot, it's really a production problem with a very specific mechanism.
| Evidence tier | Drugs |
|---|---|
| Case report (probable/definite) | Azathioprine, chloramphenicol, colchicine, cotrimoxazole, cyclophosphamide, cytarabine, 5-fluorodeoxyuridine, 5-fluorouracil, hydroxyurea, 6-mercaptopurine, methotrexate, oral contraceptives, p-aminosalicylate, phenobarbital, phenytoin, primidone, pyrimethamine, sulfasalazine, tetracycline, vinblastine |
| MedWatch postmarketing | Adalimumab, aripiprazole, carbamazepine, esomeprazole, metformin, risperidone, rivaroxaban, telaprevir |
No observational-study-tier drugs are reported for this disorder in the source appendix.
Management:stop or dose-adjust the offending drug where possible. For folate-antagonist drugs like methotrexate, leucovorin (folinic acid) rescuebypasses the blocked step in folate metabolism rather than just supplementing folic acid, which won't fix the problem if the enzyme itself is blocked.
Plain folic acid still needs the blocked enzyme (dihydrofolate reductase) to become active. Leucovorin is already the reduced, active form, so it works around a methotrexate-type block instead of running into the same wall. Mixing these up on an exam is a classic trap.
The largest table in this appendix by drug count, and the one with the widest range of mechanisms: immune complex destruction, direct marrow suppression, and, for one very specific drug, an entirely separate paradoxical thrombotic syndrome.
| Evidence tier | Drugs |
|---|---|
| Observational study evidence | Carbamazepine, oxaliplatin, phenobarbital, phenytoin, valproic acid |
| Case report (probable/definite) | Abciximab, acetaminophen, acyclovir, albendazole, aminoglutethimide, aminosalicylic acid, amiodarone, amphotericin B, ampicillin, aspirin, atorvastatin, bevacizumab, bisoprolol, capecitabine, captopril, chlorothiazide, chlorpromazine, chlorpropamide, cimetidine, ciprofloxacin, clarithromycin, clopidogrel, dabigatran, danazol, deferoxamine, diazepam, diazoxide, diclofenac, diethylstilbestrol, digoxin, ethambutol, enzalutamide, felbamate, fenofibrate, fluconazole, fondaparinux, gabapentin, gold salts, haloperidol, heparin, hydrochlorothiazide, ibuprofen, inamrinone, indinavir, indomethacin, interferon alfa-2b, isoniazid, isotretinoin, itraconazole, levamisole, levetiracetam, levofloxacin, linezolid, lithium, low-molecular-weight heparins, lurasidone, MMR vaccine, meclofenamate, mesalamine, methyldopa, minoxidil, morphine, moxifloxacin, nalidixic acid, naphazoline, naproxen, nitroglycerin, octreotide, olmesartan, oseltamivir, oxacillin, p-aminosalicylic acid, pantoprazole, penicillamine, pentamidine, pentoxifylline, piperacillin, primidone, procainamide, pyrazinamide, quinidine, quinine, ranitidine, recombinant hepatitis B vaccine, rifampin, rivaroxaban, sevoflurane, simvastatin, sirolimus, sulfasalazine, sulfonamides, sulindac, tacrolimus, tamoxifen, tolmetin, trastuzumab, trimethoprim, vancomycin |
| MedWatch postmarketing | Acarbose, adalimumab, ado-trastuzumab, alfuzosin, aliskiren, amlodipine, benazepril, bortezomib, chlorambucil, cladribine, cotrimoxazole, dalteparin, dantrolene, deferasirox, didanosine, drotrecogin alfa, efalizumab, eltrombopag, enoxaparin, epirubicin, epoprostenol, eptifibatide, filgrastim, fondaparinux, glimepiride, heparin, hydrochlorothiazide, indomethacin, iloprost, interferon beta-1a, leflunomide, losartan, montelukast, obinutuzumab, octreotide, oxcarbazepine, palivizumab, pamidronate, pemetrexed, pioglitazone, pomalidomide, propylthiouracil, quinine, raltegravir, rosiglitazone, rosuvastatin, spironolactone, sunitinib, telmisartan, torsemide, treprostinil, ursodiol |
Most drug-induced thrombocytopenia causes bleeding risk from too few platelets. Heparin-induced thrombocytopenia (HIT)is the opposite kind of dangerous: antibodies against heparin-platelet factor 4 complexes activate platelets and drive new clot formation, typically 5-10 daysafter starting heparin (or within hours on re-exposure in a previously sensitized patient). If you switch management for a falling platelet count without asking "is this HIT," you can miss a patient who's actively at risk of a limb- or life-threatening thrombosis, not just a bleed.
Management:stop the drug. For HIT specifically, stop allheparin products (including flushes) and start a non-heparin anticoagulant, since the patient still needs anticoagulation despite the low platelet count. For other drug-induced immune thrombocytopenia, platelet transfusion is reserved for active bleeding or very low counts, since transfused platelets can get destroyed by the same antibody just as fast.
Heparin(for HIT specifically), quinine and quinidine(the classic immune complex mechanism teaching example), vancomycin, and the GPIIb/IIIa inhibitors(abciximab, eptifibatide) are the names most likely to anchor an exam question or a real consult.
The specific therapy differs by disorder, but the same three moves apply almost every time.
Even when a drug is essential and alternatives are limited, deciding whether to re-challenge (or attempt a desensitization-style approach) belongs to the treating specialist, weighing the mechanism, the severity of the original reaction, and whether any alternative exists. Your job on the front line is to recognize the reaction, stop the drug, and make sure the decision gets made deliberately instead of by accident.
| Parameter | When | Watching for |
|---|---|---|
| CBC with differential | Baseline before starting a high-risk drug, then per that drug's specific schedule (e.g., clozapine's mandatory ANC checks) | Any downward trend in one or more cell lines, not just a single abnormal value |
| Reticulocyte count | Whenever a cytopenia is identified | Low: production problem (aplastic, megaloblastic). Rising: appropriate marrow response to active destruction (hemolytic) |
| Signs/symptoms of infection | Continuously in any patient on a drug linked to agranulocytosis or aplastic anemia | Fever, sore throat, mouth ulcers, which in a neutropenic patient means treat now, don't wait on labs |
| Bleeding/bruising | Continuously in any patient on a drug linked to thrombocytopenia or aplastic anemia | Petechiae, purpura, mucosal bleeding, or (for heparin specifically) new clot rather than bleeding |
| Jaundice, dark urine | Continuously in any patient on a drug linked to hemolytic anemia | Signs of active hemolysis; pair with LDH, haptoglobin, indirect bilirubin if suspected |
| G6PD status | Before starting a known oxidative-stress drug in an at-risk population, or a few weeks after an unexplained hemolytic episode | Deficiency, which changes future drug selection permanently, not just for this one course |