Clinical features and outcomes of patients with non-erythrodermic mycosis fungoides with high blood tumor burden

TO THE EDITOR:

Cutaneous T-cell lymphomas (CTCLs) represent a heterogeneous group of mature T-cell malignancies with diverse clinical behavior, biology, and outcomes.1, 2, 3 Among these, mycosis fungoides (MF) and Sézary syndrome (SS) comprise the majority of cases, although, they are often categorized together despite being distinct clinicopathologic entities with separate cells of origin.2, 3, 4, 5 Early-stage MF is commonly characterized by skin-limited patches and plaques with an indolent course, whereas SS is defined by erythroderma and a high leukemic burden of malignant T-cells in peripheral blood, conferring a poor prognosis.^3,6, 7, 8

Both MF and SS are clonal malignancies of CD4-positive T-cells that share overlapping immunophenotypic features. Therefore, disease classification relies heavily on the pattern of skin and blood involvement captured by the Tumor, Node, Metastsis, Blood (TNMB) staging system.^4,6 High leukemic burden and erythroderma—classified as B2 and T4 disease, respectively— with matching T-cell clones in skin and blood, comprise the diagnostic criteria for classic SS (cSS).^4,9

A growing body of evidence has identified a unique subset of patients who present with high blood tumor burden meeting B2 criteria but lacking the erythroderma that defines classic SS.^10,11 These patients, labeled here as non-erythrodermic SS (neSS), challenge traditional staging paradigms and raise important questions regarding disease biology, clinical behavior, and optimal management. To address this gap, we sought to characterize the clinical features and outcomes of a large cohort of neSS and cSS patients.

A retrospective study of neSS and cSS patients diagnosed after 1 January 2010 was conducted at the Abramson Cancer Center, University of Pennsylvania. Patients with B2 disease with or without T4 disease at diagnosis were identified. Skin disease was diagnosed and confirmed with dermatopathology review requiring a clonally matched T-cell population with the leukemic disease. B2 disease was defined as positive blood T-cell receptor clonality and Sézary cell (CD4⁺/CD7^˗ or CD4⁺/CD26^˗) count ≥1000 cells per μL by flow cytometry. T4 was defined as erythroderma involving ≥80% body surface area (BSA). Patients with HTLV-I/II and B2 disease utilizing only lymphocyte percentages /elevated CD4:CD8 ratio criteria were excluded as per modified International Society for Cutaneous Lymphoma/United States Cutaneous Lymphoma Consortium/European Organization for Research and Treatment of Cancer (ISCLC/USCLC/EORTC) classification.⁴ T0/B2 patients were diagnosed based on Sézary immunophenotype and exclusion of other diagnoses (eg, T-cell prolymphocytic leukemia [T-PLL], T-cell large granular lymphocytic leukemia [T-LGLL], adult T-cell leukemia/lymphoma [ATLL]). T0/B2 stage patients with disease-associated variants (DAVs) commonly seen in T-follicular helper cell lymphomas (IDH2, TET2, DNMT3A, RHOA or PLCG1) were excluded.12, 13, 14 Fisher’s exact test and Wilcoxon rank-sum test were used for statistical significance.

Two cohorts of patients with B2 disease were identified: 21 patients without erythroderma (neSS) and 79 patients with cSS, that is T4/B2 stage at diagnosis (Table 1). There were no significant differences in median age (70 and 71.5 years), race (76% and 74% White, 24% and 25% Black) or absolute Sézary cell count at diagnosis (2225 vs 2812 cells per μL) for neSS vs cSS, respectively. A significantly greater proportion of neSS patients were female (76.2% vs 44.3%, P < .01); significantly fewer had N2/3 nodal disease at diagnosis (4.8% vs 43.4%, P < .01). Additionally, significantly more cSS patients had an elevated lactate dehydrogenase at diagnosis compared to neSS (50% vs 94.6%, P < .01). One neSS patient was incidentally noted to have visceral disease pathologically confirmed in the liver (M1). No visceral disease was observed in the cSS patient cohort.

Table 1.

Clinical characteristics, DAVs and treatments received of patients in neSS and SS cohorts

	neSS (n = 21)	SS (n = 79)	P
Median age (range), y	70 (45-85)	71.5 (34-91)	.65
Race
White	76.2% (16/21)	74.6% (53/71)	1
Black	23.8% (5/21)	25.4% (18/71)
Sex
Female	76.2% (16/21)	44.3% (35/79)	.01
Male	23.8% (5/21)	55.7% (44/79)
Median Sézary cell count, cells per μL (range)	2 225 (1000-13 043)	2 812 (1 000-65 765)	.27
Elevated LDH	50% (9/18)	94.5% (69/73)	.01
Lymph node disease (N2/N3)	4.8% (1/21)	43.4% (23/53)	.01
Visceral disease (M)	4.8% (1/21)	0% (0/53)	.28
DAVs	P53 33.3% (3/9)	P53 29.4% (5/17)
	PTEN 11.1% (1/9)	PLCG1 11.8% (2/17)
	DNMT3A 11.1% (1/9)	STAT3 11.8% (2/17)
	EGR2 11.1% (1/9)	TET2 5.9% (1/17)
		KMT2A 5.9% (1/17)
Median systemic treatments (range)	3.5 (1-6)	3 (0-12)
ECP	76.2% (16/21)	59.5% (47/79)
Bexarotene	52.4% (11/21)	59.5% (47/79)
IFN	38.1% (8/21)	62% (49/79)
Mogamulizumab	38.1% (8/21)	48.1% (38/79)
Romidepsin	19% (4/21)	22.8% (18/79)
Brentuximab vedotin	14.3% (3/21)	12.7% (10/79)
Pembrolizumab	14.3% (3/21)	11.4% (9/79)
Chemotherapy	4.8% (1/21)	13.9% (11/79)
Methotrexate	4.8% (1/21)	10.1% (8/79)
Clinical trial	4.8% (1/21)	5.1% (4/79)
Allogenic HSCT	0%	8.9% (7/79)
Vorinostat	0%	7.6% (6/79)
Alemtuzumab	0%	12.7% (10/79)
Pralatrexate	0%	3.8% (3/79)
Develop erythroderma
Never	85.7% (18/21)
Yes	14.3% (3/21)
Median time to erythroderma, n = 3 (range), y	1.9 (1.3-4)
Median follow-up, total years, n = 21 (range)	5 (0.8-6.9)
Median follow-up of never erythroderma years, n = 18 (range)	4.5 (0.8-6)
Diffuse pruritus (Dx)	55% (11/20)
Skin disease (Dx)
T0	14.3% (3/21)
T1	33.3% (7/21)
T2	52.4% (11/21)

Values in bold denote significant difference (P < .05).

Wilcoxon rank-sum test and Fisher’s exact test used for statistical significance.

Dx, at diagnosis; HSCT, hematopoietic stem cell transplant; LDH, lactate dehydrogenase.

Of the neSS patients, 3 had no skin involvement (T0), 6 had patch/plaque disease involving <10% BSA (T1a/b), and 13 had patch/plaque disease involving >10% BSA (T2a/b) at diagnosis. Pruritus was reported in 55% (11/20) of neSS patients at diagnosis. Three neSS patients later developed erythroderma (T4), after a median time of 1.9 years to T4 disease from B2 diagnosis. Median follow-up for these 3 patients was similar to the 18 patients who never developed erythroderma (5 and 4.5 years, respectively). One of the patients diagnosed with T0 disease at B2 diagnosis later developed T2 disease ∼5 years after the original B2 leukemic presentation.

Interrogation of 9 neSS patient samples for cancer-associated mutations showed known DAV involving TP53 (n = 3), PTEN (n = 1), DNMT3A (n = 1) and EGR2 (n = 1). Of the three patients who presented with T0 disease at B2 diagnosis, DAVs were assessed in all 3 patients: 2 had TP53 mutations, the third did not have DAVs. Cancer associated mutations were assessed in 17 cSS patient samples with DAVs observed in TP53 (n = 5), PCLG1 (n = 2), STAT3 (n = 2), TET2 (n = 1) and KMT2A (n = 1).

The median number of systemic treatments for neSS patients was similar to cSS patients (3.5 vs 3) (Table 1). Among the neSS patient cohort, systemic treatments included extracorporeal photopheresis (ECP) (n = 16), bexarotene (n = 11), interferon (IFN) (n = 8), mogamulizumab (n = 8), romidepsin (n = 4), brentuximab vedotin (n = 3), pembrolizumab (n = 3), chemotherapy (n = 1), methotrexate (n = 1) and clinical trial (n = 1). ECP was the most common first-line therapy. No neSS patients underwent hematopoietic stem cell transplant. Among the cSS patients, IFN (n = 49), ECP (n = 47), bexarotene (n = 47), mogamulizumab (n = 38), and romidepsin (n = 18) were the most common systemic treatments, similar to the neSS cohort.

Four neSS patients underwent surveillance or skin-directed treatment without systemic therapy. Radiation was commonly used, with 4 neSS patients receiving localized radiation, 4 receiving total skin electron beam therapy, and 4 receiving narrow-band UV phototherapy. Of the 3 neSS patients who later developed erythroderma, treatments prior to developing T4 disease included ECP (n = 3), IFN (n = 3), bexarotene (n = 3), brentuximab vedotin (n = 2) and mogamulizumab (n = 1). The patient who received mogamulizumab had the longest interval between B2 and T4 diagnosis: 4 years.

When compared to the cSS patients, neSS patients had significantly better outcomes. Two-year overall survival (OS) for neSS vs cSS was 100% (18/18) and 74% (54/73) (P = .0275); 5-year OS was 100% (11/11) vs 34.5% (19/55) (P < .0001), respectively (Figure 1). When excluding neSS and cSS patients with N2/N3 lymph node involvement at diagnosis, 2-year OS was 100% (18/18) and 77% (20/26) (P = .066) and 5-year OS was 100% (10/10) and 53% (10/19) (P < .02), respectively.

Figure 1.

Kaplan Meier curve of neSS and SS cohorts. Two-year and 5-year OS rates between cohorts shown. Fisher’s exact test used for statistical significance.

Our study highlights neSS as a clinically distinct subset of CTCL characterized by high blood tumor burden in the absence of erythroderma. Compared with cSS, patients in this cohort demonstrated notable differences in clinical features, including a higher proportion of women, lower rates of elevated lactate dehydrogenase at diagnosis, and less frequent advanced nodal involvement at diagnosis. Importantly, despite sharing leukemic B2 disease, patients with non-erythrodermic disease experienced significantly higher OS at both two and five years. This difference remained statistically significant at five years even when patients with N2/N3 lymph node involvement at diagnosis were excluded.

These findings suggest that leukemic involvement alone does not uniformly confer the poor prognosis traditionally associated with SS. While the majority of non-erythrodermic patients required systemic therapy, only a minority ultimately progressed to erythroderma. This observation raises important questions regarding disease evolution, optimal timing of intervention, and whether current staging frameworks fully capture clinically meaningful risk in this population.

Our understanding of the factors driving these differences is limited. Possible underlying mechanisms that may explain the different clinicopathological behavior include T-cell mutations, host-based differences in immune microenvironment, and variations in immune signaling pathways, need to be explored. Of note, cutaneous infiltrate in CTCL is heterogenous and may include reactive immune cells responding to the tumor microenvironment. Further work is also needed to define optimal management strategies, including treatment selection and sequencing. Ultimately, improved characterization of this unique cohort may inform more precise classification and personalized therapeutic approaches for patients with advanced CTCL.

This study was Institutional Review Board (IRB) approved at the University of Pennsylvania and conducted in accordance with the Declaration of Helsinki.

Conflict-of-interest disclosure: C.J.T. Acrotech, Versed, Kyowa Kirin and Electra Therapeutics consulting; DL-Genentech, Roche, Kite, Ipsen, ADC, AbbVie consulting; SDN-Eli Lilly consulting. J.S. Bristol Myers Squibb (BMS), Seagen, Adaptive, Genmab, Kite, AztraZeneca, Incyte, CRISPR, Merck, Genmab, ATARA, Adaptive consulting. E.C. Nurix, CARGO, AbbVie, BeiGene, AztraZeneca, Genmab, Genentech consulting. S.J.S. Acerta, Nordic Nanovector, Genentech, Celgene, Merck, Celgene, Janssen, Gilead consulting. E.T. BMS, Sanofi, Genentech, Genmab consulting. M.R. Acera Bio, AbClon, Vittoria, ONI, GLG, Biocartis, CURIOX, ModeX, Beckman-Coulter consulting. S.K.B. Daiichi Sankyo, ONO, Citius, Secura Bio, Janssen, Vittoria, Acrotech, Kyowa Kirin, Kymera consulting. The remaining authors declare no competing financial interests.