Clinical features and prognostic outcomes of angioimmunoblastic T cell lymphoma in an Asian multicenter study

Abstract

In our Asian multicenter retrospective study, we investigated the clinical prognostic factors affecting the outcomes of AITL patients and identified a novel prognostic index relevant in the Asian context. In our 174-patient cohort, the median PFS and OS was 1.8 years and 5.6 years respectively. Age > 60, bone marrow involvement, total white cell count >12 × 10⁹/L and raised serum lactate dehydrogenase were associated with poorer PFS and OS in multivariate analyses. This allowed for a prognostic index (AITL-PI) differentiating patients into low (0-1 factors, n = 64), moderate (2 factors, n = 59) and high-risk (3-4 factors, n = 49) subgroups with 5-year OS of 84.0%, 44.0% and 28.0% respectively (p < 0.0001). POD24 proved to be strongly prognostic (5-year OS 24% vs 89%, p < 0.0001). Exploratory gene expression studies were performed and disparate immune cell profiles and cell signaling signatures were seen in the low risk group as compared to the intermediate and high risk groups.

Keywords:

• Angioimmunoblastic T cell lymphoma
• T follicular helper cell lymphoma
• prognostication
• gene expression analysis
• molecular genetics

Introduction

Angioimmunoblastic T cell Lymphoma (AITL) is a subtype of peripheral T cell lymphoma that was initially described to be of ‘many names’ and ‘many faces.’ [1] It was characterized as an aggressive lymphoma associated with a myriad of inflammatory and immune conditions, typically seen in the older population and presenting at advanced stages with a significant chance of bone marrow involvement [2, 3].

Under the 2016 World Health Organization (WHO) classification, AITL belongs to an umbrella group of nodal lymphomas of T follicular helper (Tfh) cell origin. Lymphomas in this group share similar molecular abnormalities related to the Tfh cell of origin, including that of RHOA, TET2, DNMT3A and IDH2 mutations [4]. On histology, AITL is defined by a partial or total effacement of the nodal architecture with a distinctive proliferation of branching high endothelial venules and surrounding CD21 positive follicular dendritic network [5]. Large B immunoblasts or multinucleated Hodgkin/Reed-Sternberg (HRS) – like cells may be seen [6]. Immunohistochemical assessment of AITL will find the neoplastic cells to be T-cell receptor α/β T cells expressing pan-T cell antigens CD2, CD3, CD5, CD7 with frequent aberrant loss or downregulation of one or more of these markers [7]. In addition, the cells are usually CD4+ and are also positive for Tfh-antigens including cell surface antigens CD10, PD-1, ICOS, CD200 and cytoplasmic SAP; transcription factor BCL6, c-MAF; chemokine CXCL13 and its receptor CXCR5 [8–13]. Of the multiple TFH markers, PD1 and ICOS are the most sensitive whereas CXCL13 and CD10 are less sensitive but more specific [14].

AITL prognosis is generally considered to be poor, with 5-year overall survival (OS) reported between 32-41% from 2 French data sets, 44% from the International T-cell Lymphoma Project (TCP) and 41% from a retrospective Japanese dataset [3,15–18]. Various prognostic indices have been established, with the International Prognostic Index (IPI) and the Prognostic index for T cell lymphoma (PIT) being still commonly used today. However, the IPI was designed for aggressive non-Hodgkin lymphomas of which Diffuse large B cell lymphomas (DLBCL) make up the bulk while PIT was designed mainly for PTCL, not otherwise specified (PTLC-NOS) [19, 20]. As such AITL specific models were developed from different datasets. Using a combination of age, performance status, extranodal involvement, B symptoms and platelet count, the Prognostic Index for AITL (PIAI) derived from 243 AITL patients from the TCP was able to differentiate patients into low (0-1 factors) and high-risk (2-5 factors) subgroups [2]. More recently, the TCP dataset was expanded and in the 282 patient cohort recruited from over 13 countries, the authors reported a novel AITL score comprising of age, ECOG performance status, and 2 biochemical markers composed of serum C-reactive protein (CRP) and serum beta 2-microglobulin levels. The AITL score stratified patients to low, intermediate and high-risk groups with 5-year OS estimates of 65%, 54% and 21% respectively. They also found that progression of disease within 24 months (POD24) was strongly prognostic [18].

In our Asian multicenter study, we studied the clinical prognostic factors affecting the outcomes of our AITL patients and aimed to develop a prognostic index that would be relevant to our population. We also aimed to identify if patients in the different prognostic groups would have differing gene expression profiles and immune signatures.

Patients and methods

Study cohort

Patients diagnosed with AITL and seen at the National Cancer Center Singapore, Singapore General Hospital and National University Cancer Institute, Singapore between June 1999 and December 2019, were retrospectively analyzed. A total of 174 patients were included in the final analysis. All data was obtained at the time of diagnosis or subsequent follow-up. The research study was carried out with approval from the SingHealth Centralized Institutional Review Board. Participants and/or their legal guardians provided informed consent for their data to be used in this research.

Demographic and clinicopathological analysis

Demographical information available included age, sex and ethnicity. Clinical characteristics of each patient including presence of B-symptoms, Eastern Cooperative Oncology Group (ECOG) performance status, Ann Arbor staging, bone marrow involvement, number of extranodal sites, chemotherapeutic regimen and response to treatment were included in the study. Biochemical and hematological information such as serum lactate dehydrogenase (LDH) level, peripheral blood leucocyte or total white (TW) count and peripheral blood platelet (PLT) count were included in the analysis. Cases of diagnostic difficulty were discussed in detail at a histo-morphology meeting and tumor board before a consensus diagnosis was made to reduce the rate of discordant diagnosis. Each patient’s International Prognostic Index (IPI) score, Prognostic Index for T-cell lymphoma (PIT) score, and PIAI (Prognostic index for AITL) score was charted for comparison of outcomes.

Study endpoints

The outcomes of interest in this study are overall survival (OS) and progression-free survival (PFS). OS was calculated from the date of diagnosis up to the date of death from any cause, or was censored at the date of last follow-up for survivors. PFS was defined as the time elapsed between the date of diagnosis to the date of relapse, progression, or death from any cause. Outcomes between patients with progression within 24 months (POD24) after diagnosis and those without POD24 were also compared.

NanoString gene expression profiling

We used the NanoString PanCancer IO360 panel (NanoString Technologies, Seattle, WA, USA) to interrogate gene expression on FFPE tissue, following manufacturer’s protocol using the nCounter platform. Briefly, RNA was extracted from five 10 μm sections on all samples with adequate tumor tissue available and analyzed using the 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). After excluding samples with suboptimal RNA integrity and content, the remaining samples were included in the nCounter analysis. The final set of data available (n = 23) was analyzed on the nSolver 4.0 Advanced Analysis module using default settings to derive differentially-expressed genes, pathway scores, and cell type scores.

Statistical analysis

For each individual clinicopathological parameter, Kaplan-Meier survival curves were plotted to estimate survival. The log-rank test was then used to determine hazard ratio (HR), the corresponding 95% confidence intervals of mortality and the p-values for each individual clinicopathological characteristic. Clinicopathological parameters found to be significant on univariate analysis using a two-sided test with significance level of 0.05 were identified. Subsequently, parameters with significance level of <0.10 were used in the generation of Multivariate Cox regression models via a backward procedure to test for independence of significant factors. Covariates identified to have an independent prognostic effect on OS were used to create a prognostic index (AITL-PI) that was relevant to our population. The performance of AITL-PI was evaluated using the Area under the curve (AUC)/concordance index (C-index). Statistical analysis was carried out using methods as previously described. All tests were performed using MedCalc statistical software for Windows version 19.0.4 (MedCalc Software, Ostend, Belgium).

Results

Patient demographics and clinicopathological characteristics

A total of 174 patients were included in this study. The median age of diagnosis was 62.3 years (range: 27.1 to 88.3 years). Ninety-nine (56.9%) were male and 75 (43.1%) were female. The ethnic distribution of our patient population was reflective of our Singaporean population with Chinese, Malay and Indians contributing 72.4%, 10.3% and 4.0% respectively. The remaining 12.6% comprises of patients from other ethnicities, largely patients from Asian countries who traveled to Singapore to seek treatment. About half (n = 93, 53.5%) the patients had B symptoms and the majority (n = 161, 92.5%) of the patients had good performance status at ECOG 0-1. Roughly three-quarters (n = 136, 78.2%) of the patients had presented at advanced stages (Ann-arbor stage 3-4) and about one-thirds of the patients had bone marrow involvement on diagnosis (n = 57, 32.8%).

Serum LDH was elevated (serum LDH > ULN) in 73.0% while blood total white cell count was elevated beyond 12 × 10⁹/L (TW > 12 × 10⁹/L) in 18.4% of the patients. Platelet counts were lower than 150,000/mm³ (PLT ≤ 150,000/mm³) in 15.5% of the patients (Table 1). PIT, IPI and PIAI score distribution in this population will be described later.

Table 1. Clinical and demographic characteristics of the overall cohort.

Characteristic	N (%)
Total	174 (100)
Age (years)
Median (range)	62.3 (27.1, 88.3)
>60	102 (58.6)
≤60	72 (41.4)
Sex
Male	99 (56.9)
Female	75 (43.1)
Ethnicity
Chinese	126 (72.4)
Malay	18 (10.3)
Indian	7 (4.0)
Others	22 (12.6)
B-symptoms
Absent	81 (46.6)
Present	93 (53.4)
ECOG performance status
0-1	161 (92.5)
2-4	9 (5.2)
Ann Arbor stage
1	14 (8.0)
2	21 (12.1)
3	65 (37.4)
4	71 (40.8)
Extra-nodal site > 1
No	148 (85.1)
Yes	22 (12.6)
Bone marrow involvement
No	101 (58.0)
Yes	57 (32.8)
Raised serum LDH level
No	42 (24.1)
Yes	127 (73.0)
TW > 12 × 10^9/L
No	139 (79.9)
Yes	32 (18.4)
PLT <150,000/mm^3
No	143 (82.2)
Yes	27 (15.5)

ECOG: Eastern cooperative oncology group; LDH: Lactate dehydrogenase; IPI: International Prognostic Index; TW: Total white cell count; PLT: Platelets count.

Missing data: 4 (2.3%) ECOG status, 3 (1.7%) stage, 4 (2.3%) extranodal sites, 16 (9.2%) bone marrow, 5 (2.9%) serum LDH level, 3 (1.7%) TW level, 4 (2.3%) PLT level, 4 (2.3%) IPI score, 3 (1.7%) PIT score, 2 (1.1%) PIAI score.

Majority of the patients received anthracycline based combinational chemotherapy for first line treatment. One hundred and nine (62.6%) patients received CHOP-like chemotherapy including 23 patients who had addition of etoposide to CHOP. Of the remaining 65 patients, 14 had non-anthracycline, cyclophosphamide-based therapy (either CEOP, CVP or CEPP), five received ifosfamide based therapy (either GIFOX or ICE), four received cytarabine based therapy (ESHAP), one received a non-chemotherapy combination of romidepsin and azacytidine, 11 received minimal treatment (either steroids, cyclosporine A, thalidomide or a combination of the former three) and 16 did not receive any treatment at all. Fourteen patients had unknown treatment regimes. On relapse, second line treatment included ifosfamide based chemotherapy in 12 patients, gemcitabine based chemotherapy in 11 patients, cytarabine based therapy in five patients and CHOP or cyclophosphamide backbone therapy in seven patients. Fourteen patients received non-chemotherapy second line agents including romidepsin, pabinostat, brentuximab and azacytidine. Only 12 patients had autologous stem cell transplant and five patients had allogeneic stem cell transplant on relapse in this cohort.

Survival analyses

The median follow-up duration for this group of patients was 20.4 months (range 0-178.8 months). The 5-year OS and PFS was 54% and 42% respectively. Median OS and PFS were 5.6 years and 1.8 years respectively (Figure 1). In univariate analyses, Age ≥60 years (HR 2.56, 95% CI 1.60–4.10, p = 0.0001), presence of B symptoms (HR 2.15, 95% CI 1.35–3.42, p = 0.0012), ECOG status 2-4 (HR 17.56, 95% CI 4.46 − 69.07, p = 0.001), Ann-Arbor stage 3-4 (HR 1.87, 95% CI 1.04 − 3.37), Bone marrow involvement (HR 2.33 95% CI 1.39–3.92), serum LDH > ULN (HR 2.63, 95% CI 1.58–4.37), TW > 12 × 10⁹/L (HR 3.63 95% CI 1.76–7.50) and PLT ≤ 150,000/mm³ (HR 2.15, 95% CI 1.04–4.44) were significantly correlated with worse OS. In terms of PFS, Age ≥60 years, presence of B symptoms, ECOG status 2-4, Ann-Arbor stage 3-4, Bone marrow involvement, serum LDH > ULN and TW > 12 × 10⁹/L were significantly correlated with inferior PFS (Table 2).

Figure 1. Survival probabilities of the overall AITL cohort. (A) Kaplan-Meier curve of overall survival of our AITL cohort. (B) AITL survival stratified according to IPI risk groups. (C) AITL survival stratified according to PIT risk groups. (D) AITL survival stratified according to PIAI risk groups.

Table 2. Univariate analysis for overall survival and progression free survival.

Characteristic	Overall survival		Progression free survival
	p value	HR (95% CI)	p value	HR (95% CI)
Age > 60 years	0.0001	2.56 (1.60–4.10)	0.0007	2.02 (1.34–3.04)
Presence of B symptoms	0.0012	2.15 (1.35–3.42)	0.0037	1.81 (1.21–2.72)
ECOG > 1	0.0001	17.56 (4.46–69.07)	< 0.0001	138.85 (28.12–685.45)
Ann Arbor stage 3 – 4	0.0351	1.87 (1.04–3.37)	0.0158	1.83 (1.12–2.99)
Extranodal site > 1	0.4409	1.30 (0.66–2.57)	0.4303	1.27 (0.66–2.26)
Bone marrow involvement	0.0013	2.33 (1.40–3.92)	0.0004	2.31 (1.45–3.68)
LDH > ULN	0.0002	2.63 (1.58–4.37)	<0.0001	2.61 (1.68–4.05)
TW > 12 × 10^9/L	0.0087	2.44 (1.25–4.76)	0.0019	2.51 (1.40–4.50)
PLT < 150,000/mm^3	0.0383	2.15 (1.04–4.44)	0.3117	1.37 (0.74–2.53

HR: hazard ratio, CI: confidence interval, ECOG: Eastern cooperative oncology group, LDH: Lactate dehydrogenase, ULN: Upper limit of normal, TW: Total white cell count, PLT: Platelets count.

Derivation of a novel Asian AITL index

In our multicenter Asian AITL cohort, a multivariate analysis adjusted for significant clinicopathological parameters for OS only four factors returned with an independently associated prognostic value and they were Age ≥60 years, Bone marrow Involvement, Total white cell count greater than 12 × 10^9/L and serum LDH greater than upper limit of normal (Table 3 and Suppl. Figure 1). These same factors also retained its prognostic value in terms of PFS in the multivariate analysis (Table 3). Combining these four factors, we derived a prognostic index (AITL-PI) which stratifies patients into low (0-1 factors, n = 64, 37.2%), moderate (2 factors, n = 59, 34.3%) and high-risk (3-4 factors, n = 49, 28.5%) subgroups (Figure 2(A)).

Figure 2. Survival probabilities of patients. Overall survival as classified by (A) AITL prognostic index (B) with or without POD24.

Table 3. Multivariate analysis for overall survival and progression free survival.

Characteristic*	Overall survival		Progression free survival
	p value	HR (95% CI)	p value	HR (95% CI)
Age > 60 years	0.0014	2.50 (1.42–4.41)	0.0137	1.83 (1.13–2.97)
Bone marrow involvement	0.0145	1.87 (1.13–3.09)	0.0045	1.89 (1.21–2.93)
TW > 12 × 10^9/L	0.0324	1.88 (1.05–3.37)	0.0099	1.93 (1.17–3.19)
LDH > ULN	0.0211	2.58 (1.15–5.78)	0.0026	2.98 (1.46–6.08)

OS AUC/C-index: 0.722 (95% CI: 0.644–0.791).

PFS AUC/C-index: 0.772 (95%CI: 0.697–0.835).

*Parameters with significance level of < 0.10 in univariate analyses were used in the generation of Multivariate Cox regression models via a backward procedure. ECOG status was not included in view of small numbers of patients with ECOG >1 (n = 9).

In comparison to previously established prognostic indices, the AITL-PI was able to differentiate our cohort into the three balanced risk groups with disparate survival outcomes, Patients with low, intermediate and high-risk AITL-PI scores had 5-year PFS estimates of 72%, 32.5% and 16.5% respectively (p < 0.0001), and 5-year OS estimates of 84.0%, 44.0% and 28.0% respectively (p < 0.0001). The AUC/C-index for multivariate analysis for OS was 0.722 (95% CI: 0.644 − 0.791) and the corresponding values for multivariate analysis for PFS was 0.772 (95%CI: 0.697 − 0.835). Stratification of patients using the IPI, PIT and PIAI score also returned with significant differences in PFS and OS in their corresponding low and high risk groups (Table 4).

Table 4. Comparison of AITL prognostic index with other established prognostic indices.

Prognostic Index	IPI		PIT		PIAI		AITL-PI
Risk group (score)	Low (0-2)	High (3-5)	Low (0-1)	High (2-4)	Low (0-1)	High (2-5)	Low (0-1)	Int. (2)	High (3-4)
N (%)	92 (54.1)	78 (45.9)	71 (41.5)	100 (58.5)	95 (55.2)	77 (44.8)	64 (37.2)	59 (34.3)	49 (28.5)
Median PFS	7.48	0.9	NR	0.84	11.8	0.9	NR	0.9	0.5
5-year PFS (%)	60.5	19.0	66.5	25.0	56.5	24.0	70.0	32.5	16.5
Median OS (years)	14.7	1.7	NR	1.7	14.7	1.7	NR	3.0	1.4
5-year OS (%)	76.0	27.5	82.0	34.5	72.0	32.0	84.0	44.0	28.0
p-value	<0.0001		<0.0001		<0.0001		<0.0001

IPI: International prognostic index; PIT: Prognostic index for T-cell Lymphoma; PIAI: Prognostic Index for AITL; PFS: Progression free survival; OS: Overall survival; NR: Not reached.

Validation of POD 24

One hundred and sixteen (66.7%) patients had progression of disease within 24 months (POD24) whereas 58 (33.3%) patients were without POD24. In patients with POD24, their 5-year OS was 24% as compared to 89% in those without POD24 (p < 0.0001). Median OS was 1.8 years vs 14.7 years respectively (Figure 2(B)).

Immune profiling

A total of 23 archival tissue samples were selected for gene expression profiling using the NanoString PanCancer IO 360 panel. (Raw data available in supplement) The patients tissue samples were stratified by their AITL-PI scores, with low risk (n = 6), moderate risk (n = 9) and high risk (n = 8) patients. In this exploratory analysis, cell type profiling revealed a higher trend of cytotoxic cell profiles in the intermediate and high risk groups whereas a neutrophilic, T-regulatory and Th-1 cell profile trended higher in the low risk groups (Figure 3(A)). In addition, analysis of immune and oncogenic signaling pathway signatures found that the low risk group were more active in the myeloid compartment, WNT signaling, cytokine and chemokine signaling, TGF-$b$ signaling and hedgehog signaling pathways. In the intermediate and high-risk groups, cytotoxicity signatures, interferon signaling and lymphoid compartment signatures returned with higher pathway scores (Figure 3(B)).

Figure 3. Gene expression analysis of patients. (A) Graphs showing gene expression profiling of immune cell types expressed in their raw scores in the AITL cohort stratified according to their risk groups using AITL Prognostic index and (B) Graph showing gene expression analysis of immune cell signaling pathway scores in the AITL cohort stratified according to their risk groups using AITL Prognostic index.

Discussion

AITL prognosis has been regarded as poor since it was first understood to be a neoplastic disease. Between 1990-2002, the 5 year reported OS for AITL was 32% in the International T-cell lymphoma project (TCP) where AITL took up 18.5% of the 1314 strong cohort [15]. This was subsequently updated in the recent AITL prognostic score published by the same group where the reported 5 year OS was 44% [18]. In our Asian multicenter study with a total of 174 AITL patients, the overall 5-year OS was notably slightly higher at 53%. In terms of demographics, what was consistent in our study was that we had a majority (58.6%) of our patients being older than 60 (median age 62.3) and there was a slight male predominance (1.32:1). The patients presented in advanced stages and B symptoms were present in about half of our patients. However, some obvious differences in the patient demographic profiles include that of the stage and ECOG performance status at presentation, the proportion with bone marrow involvement and proportion with serum LDH elevation at presentation. In the TCP, 90% of the patients presented at Ann Arbor stages 3-4 whereas about 78% in our population presented at similar stages. Likewise, 31% of the TCP population had ECOG performance status of 2 or more but this group of patients only made up 5.2% of our cohort. It is uncertain if this difference in the ECOG performance status of the two groups is a reflection of the difference in the stage of presentation or a nature of the different populations’ perspective on baseline functional status. Interestingly, despite a lower proportion of patients with stage 3-4 disease, we had a higher proportion of patients with bone marrow involvement at 33% compared to 13% in the TCP, and a higher proportion of patients with elevated LDH at 73% vs 58% in the TCP.

Notwithstanding the slight differences in the TCP and our cohort, when stratified according to non-Hodgkin lymphoma prognostic indices IPI, PIT and PIAI, significant differences in OS and PFS were still found in the respective low and high risk groups and this serves to validate the prognostic utility of these previously established indices in our Asian AITL cohort. Our study also validated POD24 as a prognostic factor that was described in the TCP. In our population the outcomes were again very disparate between the two groups, with the group without POD24 having a 5-year OS that was 3.7 times that of patients with POD24.

In the TCP’s novel AITL score reported based on a limited dataset of 96 patients with additional laboratory markers, baseline B2-microglobulin (B2M) and C-reactive protein (CRP) levels were included as part of the prognostic index. As these markers are not routinely sent for all patients in our local population, we sought to look for relevant markers from our patient’s available haematological and biochemical profiles. Blood total white cell count was used as a surrogate for the B2M and CRP as both reflect a raised inflammatory state and TW count was readily available in our study. Indeed in our univariate analysis, TW >12 g/L returned as a factor that retained in prognostic significance even in the multivariate analysis. In our analysis, TW >12 g/L put together with age ≥60 years, LDH > ULN and bone marrow involvement gave rise to a AITL-PI relevant to our population and simple enough to calculate in the daily clinical setting. The AITL-PI delineated our patients into three fairly well distributed risk groups with about a third of the cohort in each group but with significantly different outcomes. In the low risk group, the 5-year OS was high at 84% compared with the high risk group at only 28%. This low risk group identifies a subset of AITL with good prognosis, and veers away from conventional knowledge that AITL is a disease always defined by poor prognosis.

In our exploratory analysis of the immune-oncological signatures of AITL tumors derived from patients of the three respective AITL-PI risk groups, we found interesting signatures that were differing in the low risk versus the intermediate and high risk groups. The increasing cytotoxicity cell profiles together with a decreasing neutrophilic, T-regulatory cell and Th-1 cell profile in the high risk groups suggest a more inflammatory and cytotoxic signature in the high risk group. This was also previously reported in a gene-expression profiling study of 372 PTCL patients which included 114 AITL patients [21]. A high expression of B cell signature in AITL patients was correlated with a more favorable outcome whereas high expression of either a cytotoxic signature associated with CD 8⁺ T cells, a monocytic signature, or a p53-induced gene signature was associated with poorer outcomes.

Separately, the high pathway scores of cytokine and chemokine signaling together with TGF-$b$ signaling in the low risk group suggest that TGF-$b$ signaling plays an important role in the low risk group but less so in the moderate to high risk groups. TGF-$b$ is known to be a pleotropic cytokine exerting various effects within a cancer microenvironment. It is typically recognized as an inhibitor of T cell proliferation, activation and effector functions and in cancer, hypothesized to assist in immune evasion. Abundant TGF-$b$ in cancer micro-environments have been shown to adversely impact cancer prognosis [22]. However, in early stage cancer, this pathway has also been recognized to promote tumor suppression, cell-cycle arrest and apoptosis [23]. Although exploratory, the decreasing trend of TGF-$b$ signaling in our low to high risk cohorts suggest that it may be playing a tumor suppressive role in the low risk AITL patients.

Likewise, interferon-$\gamma$ is yet another pleiotropic cytokine frequently associated with anti-proliferative and anti-tumor mechanisms and is a major effector of immunity. However, its role in tumorigenesis has also been recently suggested [24]. In our AITL cohort, we found an increase in interferon-$\gamma$ signaling scores in the intermediate to high risk groups, paired with the above mentioned decrease in TGF-$b$ signaling scores. This suggests that the higher risk AITL tumors have developed possible mechanisms to overcome the inherent TGF-$b$ tumor suppressive functions by adopting pro-inflammatory signaling pathways.

Interestingly, Hedgehog and WNT signaling pathways returned with high scores in the low risk groups and were reduced in the intermediate to high risk groups. Both Hedgehog and WNT are evolutionarily conserved major developmental signaling pathways whose dysregulation have been in implicated in cancer progression. Through the promotion of cancer stem cells, these pathways are also thought to contribute to cancer therapy resistance [25]. It is therefore unusual that we see reduced signaling of these pathways in the higher risk AITL patients as opposed to the low risk AITL patients. Further elucidation of the various signaling and immune mechanisms for AITL pathogenesis may allow us to identify appropriate therapeutic targets for high risk AITL patients.

As our patient population spanned 20 years from 1999 to 2019, first line treatment in the cohort was varied. Although anthracycline-based chemotherapy has not been proven to be superior to other multiagent chemotherapy for PTCL, CHOP with or without etoposide is still a standard first line chemotherapy offered to most patients with AITL [26]. In our study, 63% of the cohort had CHOP or CHOP-like treatment, of which 21% (i.e. 13% of the entire cohort) received an intensified regime with addition of etoposide to CHOP. Anecdotally, other treatment strategies in elderly patients include watch and wait approach and less intensive options such as steroids as single agents, ciclosporin, thalidomide and chidamide all with reported efficacies albeit in small and limited series [27–29]. Interestingly, 16% of our cohort had either no or minimal treatment after initial diagnosis and this was either due to patient age and fitness, or patient’s personal preference. Minimal treatment in this group included oral steroids, oral ciclosporin or thalidomide in a handful of patients.

The primary limitation of our study is that this is a retrospective analysis. Although pathological and immunohistochemical data were collected, information was incomplete for some of the fields that were of interest and thus was not reported in this study. We would have liked to examine the Attygalle score [30] distribution and the proportion of CD30 positivity in our cohort. Additionally, treatment received was varied as data collection spanned over 20 years, and did not include the patients who received brentuximab as ECHELON-2 was published only in 2019 [31].

In conclusion, we have developed a simple yet effective prognostic index that is relevant to Asian AITL patients. There may be a subset of AITL patients with good prognosis that can be differentiated from the rest. Further studies and molecular analyses are being planned to better understand this subset of patients.

Author contributions

EWYC and JYC analyzed the data and drafted the manuscript; EWYC, VSY, SYO, HXK, SDM, EKYN, MLP, YHT, JC, EP, NS, MF, TT, MT, LPK, STL and JYC obtained patient data; BYL, DH, and CKO performed and interpreted the transcriptomic data; EWYC, VSY and JYC designed the study, interpreted the results, and revised the manuscript; and all authors read and approved the final version of the manuscript.

Acknowledgements

We would like to thank all subjects who have participated in this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The datasets created and analyzed during this study are available from the corresponding author at [email protected] upon reasonable request.

Additional information

Funding

This work was supported by the NCCS Cancer Fund (Research) [NCCSCF-R-YR2021-JUN-SSD1], Tanoto Foundation Professorship in Medical Oncology, New Century Foundation Limited, Ling Foundation, Singapore Ministry of Health’s National Medical Research Council Research Transition Awards [TA21jun-0005 and TA20nov-0020], Large Collaborative Grant [OFLCG18May-0028], and Collaborative centre grant [TETRAD II].