Romidepsin in peripheral and cutaneous T-cell lymphoma: mechanistic implications from clinical and correlative data

Abstract

Romidepsin is an epigenetic agent approved for the treatment of patients with cutaneous or peripheral T-cell lymphoma (CTCL and PTCL). Here we report data in all patients treated on the National Cancer Institute 1312 trial, demonstrating long-term disease control and the ability to retreat patients relapsing off-therapy. In all, 84 patients with CTCL and 47 with PTCL were enrolled. Responses occurred early, were clinically meaningful and of very long duration in some cases. Notably, patients with PTCL receiving romidepsin as third-line therapy or later had a comparable response rate (32%) of similar duration as the total population (38%). Eight patients had treatment breaks of 3.5 months to 10 years; in four of six patients, re-initiation of treatment led to clear benefit. Safety data show slightly greater haematological and constitutional toxicity in PTCL. cDNA microarray studies show unique individual gene expression profiles, minimal overlap between patients, and both induction and repression of gene expression that reversed within 24 h. These data argue against cell death occurring as a result of an epigenetics-mediated gene induction programme. Together this work supports the safety and activity of romidepsin in T-cell lymphoma, but suggests a complex mechanism of action.

Trial registration: ClinicalTrials.gov NCT00007345.

Keywords: T-cell lymphoma; chromatin; epigenetic therapy; histone deacetylase inhibitor; romidepsin.

Figure 1

Durations of response for patients with CTCL and PTCL. Durations in patients with CR are shown in dark grey, and patients with PR shown in lighter grey. *Indicates patients censored at data cut-off still in CR or censored at off study date still in response continuing romidepsin. Patients with PTCL who received romidepsin in third-line (with two or more prior chemotherapy regimens before enrollment) are indicated by the hatched bars. CTCL, cutaneous T-cell lymphoma; PTCL, peripheral T-cell lymphoma; CR, complete response; PR, partial response.

Figure 2

Graphical display of treatment durations and intervals in patients who had treatment breaks. On-treatment interval is shown in dark grey, while the hatched areas show off-treatment interval. * Indicates that response was continuing at the time of data cut-off or study discontinuation. One patient with CTCL discontinued study to receive romidepsin at home after US Food and Drug Administration approval of the agent. CTCL, cutaneous T-cell lymphoma; PTCL, peripheral T-cell lymphoma.

Figure 3

Patient with Stage IVA cutaneous T-cell lymphoma (CTCL) diagnosed 22 months prior to study entry. Prior therapies included 2 months of psoralen photochemotherapy, 3 months of denileukin diftitox, 2 months of pentostatin and 5 months of liposomal doxorubicin. She presented at age 64 years with stage IVA disease, with skin involvement, enlarged lymph nodes, and circulating malignant cells. After 2 cycles of romidepsin, this patient achieved a partial response, and a complete response was scored at the end of Cycle 9. At Cycle 20, a single small patch was noted on her sacral area that was biopsied and confirmed for CTCL cells. As per protocol, disease progression was called, but romidepsin was continued as compassionate use. Small patch disease waxed and waned over the next 12 cycles. A decision was made to stop romidepsin after a total 32 cycles, and the patient was followed without progression another 43 months. She then presented with generalized erythroderma, increased circulating Sézary cells in her blood, and enlargement of lymph nodes. She began a 6-cycle course of romidepsin that cleared blood and skin. She was again monitored for 34 months without treatment, after which increasing disease in skin and blood prompted another 6-cycle course of romidepsin, with clearing of disease. She was monitored off therapy an additional 14 months until disease required re-initiation of another course of romidepsin, completed 12 years after her original enrollment on NCI1312. This course is summarized as Patient 1 in Supplementary Table 1. C, cycle; D, day

Figure 4

Bilirubin, ALT and AST elevation following romidepsin. Patient with PTCL (subtype PTCL not otherwise specified) previously treated with cyclophosphamide, doxorubicin, vincristine and prednisone (CHOP). The patient’s medical history was notable for fever, chills and abdominal pain, and baseline symptoms included lymphopenia, elevated alkaline phosphatase, elevated ALT and AST, hypoalbuminaemia, and hypocalcaemia. The patient started treatment with romidepsin 14 mg/m² and received a total of 39 romidepsin doses in 13 cycles with a last dose of romidepsin on C13D15, when disease progression was noted. During the early cycles, ALT and AST increased markedly after the first dose of each cycle, and increases in bilirubin were almost always noted the day after romidepsin infusion. PTCL, peripheral T-cell lymphoma; ALT, alanine transaminase; AST, aspartate transaminase

Figure 5

cDNA microarray analysis performed on 0h, 4h, and 24h samples in 7 patients with circulating Sézary cells. A. Principal component analysis (PCA) plot based on the 996 most variable genes (Filtered by variance: σ/σ_max>0.3) across all samples. The 20 samples are coloured according to patient number. B. Heat-map and unsupervised hierarchical clustering based on the same 996 most variable genes as for the PCA plot. Again, we see clear clustering according to individual patients, rather than according to time. The 4h sample for Patient 7 was eliminated as an outlier. Green shading indicates reduction in gene expression and red shading indicates an increase in gene expression. C. Heat-map and one-way hierarchical clustering based on the 290 genes that differed between 4h and 0h, and between 24h and 0h identified by a paired analysis eliminating patient as a nominal factor (p<0.01, >2-fold change). As seen, changes at 4h typically reverse by 24h, and changes at 24h were seldom seen at 4h. Only 10 genes were found in common between the two time comparisons. Expanded heat maps including gene names can be found in Supplementary Figure 3. D. Bubble diagram showing the top-14 canonical pathways over-represented in the transcriptome of the 6 patients (P1-6) at times 4h and 24h post-treatment. The colour indicates the p-value (calculated by the right-tailed Fisher Exact Test*) for each particular pathway; from dark red (P<0.00001) to pale (not significant). The size of each bubble indicates the ratio, which is the number of genes in a given pathway that meet the cut-off criteria divided by the total number of genes that make up that pathway. The larger and darker a given bubble, the more significant the finding. *The details of the calculation of the P-value can be found in this whitepaper: http://www.ingenuity.com/wp-content/themes/ingenuity-qiagen/pdf/ipa/functions-pathways-pval-whitepaper.pdf E. Ingenuity Pathway Analysis identified NFκB as one of the top altered networks at the 4 h time-point. Shown are the changes in NFκB targets at 4 and 24 h. Heat-map and 2-way hierarchical clustering based on 29 NF-κB target genes that differ significantly (p<0.05) between time points. Patient is eliminated as a nominal factor. Target genes of NF-κB were obtained from the NF-κB online resource at Boston University: http://www.bu.edu/nf-kb/gene-resources/target-genes/.