Romidepsin

Romidepsin is a natural-product macrocycle containing both amide and ester linkages — a depsipeptide — with an internal disulfide bond. Once inside the cell, glutathione reduces that bond and liberates a free thiol that inserts into the catalytic pocket of class I histone deacetylases, HDAC1 and HDAC2 in particular, and coordinates the active-site zinc. Loss of deacetylase activity leaves histones hyperacetylated, chromatin open, and a set of tumour-suppressor and pro-apoptotic genes re-expressed. Malignant T cells are unusually dependent on the transcriptional programmes this disrupts, which is why activity in cutaneous T-cell lymphoma is far better than in solid tumours. Non-histone substrates such as HSP90 and tubulin are also affected, contributing to the broader cellular stress response.

Targets: Histone deacetylase 1 (HDAC1), Histone deacetylase 2 (HDAC2), Class I HDACs generally

• · 14 mg/m2 per dose. Potassium and magnesium are repleted before every dose because the cardiac repolarisation effects are amplified by low electrolytes.

Work out your exact syringe units →

• very commonNausea and vomiting— Around 50 to 60 percent; standard antiemetic prophylaxis is used.
• very commonFatigue
• very commonThrombocytopenia, neutropenia and anaemia— Usually reversible between cycles.
• commonECG T-wave and ST-segment changes— Very frequently seen and usually benign, but QT prolongation is real — hence mandatory electrolyte repletion.
• commonAnorexia and dysgeusia
• commonInfections— Including sepsis and pneumonia in an already immunocompromised population.
• uncommonTumour lysis syndrome— Higher risk in bulky disease at the first cycle.
• rareEpstein-Barr or hepatitis B reactivation— Screen hepatitis B status before starting.

• Congenital long QT syndrome or uncorrected hypokalaemia and hypomagnesaemia.
• Concurrent strong CYP3A4 inhibitors without dose adjustment.
• Pregnancy — embryo-fetal toxicity.

• conflictStrong CYP3A4 inhibitors such as ketoconazole — Substantially increases exposure; avoid or reduce dose.
• conflictStrong CYP3A4 inducers such as rifampin — Can reduce exposure enough to lose efficacy.
• cautionQT-prolonging drugs and antiarrhythmics — Additive repolarisation risk on top of romidepsin's own ECG effects.
• cautionWarfarin — Prolonged PT and elevated INR have been reported; monitor closely.

• · Serum potassium and magnesium before every dose, repleted into the normal range.
• · ECG at baseline and periodically in patients with cardiac risk or on other QT-prolonging drugs.
• · Full blood count and chemistry before each dose.
• · Uric acid and renal function during the first cycle for tumour lysis.
• · Hepatitis B serology before starting.

FDA approved for cutaneous T-cell lymphoma after at least one prior systemic therapy; the peripheral T-cell lymphoma indication was withdrawn in the US in 2021. Hospital-administered chemotherapy.

• Whittaker et al. 2010, pivotal phase 2 of romidepsin in cutaneous T-cell lymphoma, Journal of Clinical Oncology (trial)
• Bachy et al. 2022, Ro-CHOP phase 3 trial in peripheral T-cell lymphoma, Journal of Clinical Oncology (trial)
• FDA prescribing information for Istodax (label)

Romidepsin is a prodrug in the strict sense, and the activation step is the interesting part. The intact bicyclic molecule with its disulfide bridge is essentially inactive against histone deacetylases. Once inside the cell, the abundant reducing environment — glutathione at millimolar concentrations — cleaves the disulfide and liberates a free thiol on a long aliphatic arm. That thiol is a zinc-binding group, and it threads into the narrow substrate channel of a class I histone deacetylase to coordinate the catalytic zinc ion directly, which is the same fundamental chemistry as vorinostat's hydroxamic acid but with better selectivity for class I over class II enzymes. The bicyclic scaffold acts as a cap that sits at the channel entrance and provides the selectivity. This intracellular activation is why romidepsin is far more potent in cells than a simple biochemical assay on the intact molecule would suggest. Downstream, loss of HDAC1 and HDAC2 activity leaves nucleosomal histones hyperacetylated, chromatin opens, and a set of silenced genes is re-expressed — p21/CDKN1A most reliably, which arrests the cell cycle. But the histone story is only half of it. HDACs deacetylate hundreds of non-histone substrates, and the ones that matter here include HSP90, whose acetylation disables its chaperone function and destabilises client oncoproteins, alpha-tubulin, and transcription factors including STAT3 and NF-kB components. Malignant T cells appear to depend on this transcriptional programme more than most tumours do, which is the leading explanation for why HDAC inhibitors work reasonably well in cutaneous T-cell lymphoma and essentially not at all in solid tumours. The cardiac effect is also mechanistic rather than incidental: HDAC inhibition alters expression and function of cardiac ion channels, particularly hERG-mediated repolarising current, which is why T-wave and ST-segment changes are near-universal and why electrolyte repletion before every dose is mandatory rather than cautious.

The cardiac story is the one to take seriously, and the risk is almost always compound rather than single: a patient with low magnesium from diarrhoea, on an azole antifungal that both prolongs QT and inhibits CYP3A4, receiving a drug that prolongs QT, with tumour lysis pushing potassium around. Each element alone is manageable and the combination is not. The fix is boring and effective — check and replete electrolytes on the day of every dose, audit the concomitant drug list for QT prolongers, and know the hepatitis B status before the first infusion. The second failure is expectation. This drug produces a roughly one-in-three response rate in a disease that is chronic and relapsing; it is a good drug for controlling an incurable skin lymphoma, not a curative one, and the pruritus relief it gives even to non-responders is a legitimate reason to continue that gets undervalued. Third, the reconstitution: it ships with its own co-packaged diluent, a povidone, propylene glycol and ethanol vehicle, and bacteriostatic water is not a substitute. Fourth, infection risk in an already immunocompromised population is real and includes sepsis and pneumonia at meaningful rates. Fifth, the four-hour infusion is a genuine burden three times a month and it is why some patients drift off treatment that was working.

1. —Cycle 1 onward — 14 mg/m2 intravenously over 4 hours on days 1, 8 and 15 of a 28-day cycle. Body-surface-area dosed. Electrolytes repleted before every single dose.
2. —On recurrent grade 3 or higher non-haematologic toxicity, or recurrent grade 4 cytopenia — Reduce to 10 mg/m2. Hold first until toxicity resolves to grade 1 or baseline, then restart at the lower dose.
3. —On further recurrence at 10 mg/m2 — Discontinue. There is no third dose level.
4. —Moderate to severe hepatic impairment, from the first dose — Start below the standard 14 mg/m2. The exact reduction depends on the degree of impairment and on whether a strong CYP3A4 inhibitor is also on board.

• Histone deacetylase 1 (HDAC1) — Low nanomolar IC50 after intracellular reduction; commonly cited figures for HDAC1 and HDAC2 are in the tens of nanomolar range

  Zinc chelation in the catalytic pocket, histone hyperacetylation, chromatin opening and re-expression of silenced genes including p21.

• Histone deacetylase 2 (HDAC2) — Comparable to HDAC1

  Same class I mechanism. HDAC1 and HDAC2 together account for most of the antitumour effect.

• HDAC3 and HDAC8 (remaining class I enzymes) — Weaker but meaningful

  Contributes to the overall class I blockade. Romidepsin is class-I-selective, which distinguishes it from pan-HDAC inhibitors like vorinostat.

• Non-histone substrates — HSP90, alpha-tubulin, STAT3

  HSP90 hyperacetylation disables chaperone function and destabilises client oncoproteins. This is probably a larger part of the antitumour effect than the histone story suggests.

• Cardiac repolarisation, via HDAC-dependent effects on ion channel expression including hERG

  T-wave flattening and ST-segment depression in a majority of patients, plus genuine QT prolongation. The reason potassium and magnesium are repleted before every dose.

• Pivotal international phase 2 in cutaneous T-cell lymphoma Phase 2 · n=96 · 2010

  Objective response rate by an independent review committee in refractory cutaneous T-cell lymphoma. Response rate around 34 percent with a median duration of response around 15 months, and meaningful pruritus relief even in some non-responders.

• Ro-CHOP Phase 3 · n=421 · 2022

  Progression-free survival with romidepsin plus CHOP versus CHOP alone in previously untreated peripheral T-cell lymphoma. The trial did not meet its endpoint — adding romidepsin did not improve progression-free survival and increased toxicity. This failure is why the US peripheral T-cell lymphoma indication was withdrawn in 2021.

Histone hyperacetylation is measurable in peripheral blood mononuclear cells within hours of an infusion and persists for a day or more. Pruritus relief in cutaneous T-cell lymphoma is often the first clinical change and can arrive within the first cycle, sometimes before any visible skin improvement. Skin responses typically take 2 to 4 cycles, with a median time to response around 2 months. Responses that do occur tend to be durable, with a median duration around 15 months in the pivotal study. ECG changes appear during and shortly after each infusion and resolve. Cytopenias nadir between doses and recover between cycles. Nausea is immediate and infusion-related.

Electrolyte repletion is part of the drug, not an adjunct — potassium and magnesium go into the normal range before every dose, and the schedule assumes it. Antiemetic prophylaxis is standard because nausea and vomiting affect roughly half of patients. The CYP3A4 interactions are the sharp edge. Strong inhibitors — ketoconazole, itraconazole, clarithromycin, ritonavir — substantially raise exposure and therefore the cardiac and marrow toxicity, so they are avoided or the dose is reduced. Strong inducers — rifampicin, carbamazepine, phenytoin, St John's wort — can drop exposure enough to lose efficacy, and rifampicin cut the apparent volume of distribution by 52 percent in a formal study. Warfarin is a specific trap: prolonged prothrombin time and raised INR have been reported, so anyone anticoagulated with warfarin needs closer INR monitoring than usual. The additive-risk list is anything that prolongs QT — ondansetron at high dose, azoles, macrolides, fluoroquinolones, methadone, several antipsychotics — and in a lymphoma patient on antifungal and antiemetic prophylaxis, that stack assembles itself without anyone deciding to build it. Bexarotene, phototherapy and skin-directed treatments combine reasonably in cutaneous T-cell lymphoma. The one that failed is CHOP, and it failed prospectively.

Against vorinostat, the oral pan-HDAC inhibitor also approved for cutaneous T-cell lymphoma, romidepsin is more potent, more class-I-selective and more myelosuppressive, and it is intravenous where vorinostat is a tablet — so vorinostat wins on convenience and romidepsin on response rate. Against brentuximab vedotin in CD30-positive disease, brentuximab has substantially better response rates and is the preferred option when CD30 is expressed. Against mogamulizumab, the anti-CCR4 antibody, mogamulizumab performs better in blood-involved Sézary syndrome. Against belinostat in peripheral T-cell lymphoma, the class effect is similar and neither has a confirmed survival benefit. The most instructive comparison is romidepsin against itself: the same drug held two indications, and the one supported by a confirmatory phase 3 that failed was withdrawn while the one supported by durable phase 2 responses survived. That is what an honest regulatory system looks like.

Not characterised. Romidepsin is a brand-and-generic intravenous drug dosed by body surface area, delivered as three four-hour infusions per 28-day cycle in an infusion centre, and the chair time is a substantial part of the real cost. A vial price would not represent what a cycle actually costs.

• ?Volume of distribution and clearance are not published as single numeric values in the label, so both are left null. The only quantitative handle is the 52 percent reduction in apparent volume of distribution with rifampicin.
• ?The commonly quoted HDAC1 and HDAC2 IC50 values in the tens of nanomolar range come from the general literature; I did not resolve a primary source in this pass.
• ?Blood-brain barrier penetration is marked unknown. There is no good human data and the molecule's size and protein binding argue against meaningful penetration, but that is inference rather than measurement.
• ?Whether the antitumour effect is driven mainly by histone or by non-histone substrate acetylation is genuinely unresolved, and the HSP90 mechanism may be more important than the chromatin story that gives the class its name.
• ?The pivotal phase 2 enrolment of 96 and the Ro-CHOP enrolment of 421 are as commonly cited; I resolved both publication records but not participant-level detail.
• ?The mechanism linking HDAC inhibition to cardiac repolarisation changes is plausible and widely repeated but not definitively established.
• ?Protein binding is entered as 93 percent, the midpoint of the label's stated 92 to 94 percent range, because the schema takes a single number.

• Final results from a multicenter, international, pivotal study of romidepsin in refractory cutaneous T-cell lymphoma Whittaker SJ, Demierre MF, Kim EJ, et al., Journal of Clinical Oncology, 2010 · PMID 20697094

  The registration study for the surviving indication, and the source of the roughly 34 percent response rate with durable responses.

• Romidepsin Plus CHOP Versus CHOP in Patients With Previously Untreated Peripheral T-Cell Lymphoma: Results of the Ro-CHOP Phase III Study (Conducted by LYSA) Bachy E, Camus V, Thieblemont C, et al., Journal of Clinical Oncology, 2022 · PMID 34843406

  The confirmatory trial that failed, leading to withdrawal of the peripheral T-cell lymphoma indication. A clean example of accelerated approval working the way it is supposed to.

• Population pharmacokinetics of romidepsin in patients with cutaneous T-cell lymphoma and relapsed peripheral T-cell lymphoma Woo S, Gardner ER, Chen X, et al., Clinical Cancer Research, 2009 · PMID 19228751

  The dedicated population pharmacokinetic analysis, useful because the label reports so few numeric disposition parameters.

• ISTODAX (romidepsin) for injection — US prescribing information Celgene / Bristol Myers Squibb, FDA prescribing information

  Source of the 377 ng/mL Cmax, 92 to 94 percent protein binding with alpha-1-acid glycoprotein as the principal binding protein, the roughly 3-hour terminal half-life and the CYP3A4-dominant metabolism.