Effect of POD24 on the prognosis of follicular and other indolent B-cell non-hodgkin lymphomas

Highlights

Does the presence of POD24 serve as a reliable prognostic factor for adverse outcomes in all individuals diagnosed with B-cell indolent lymphoma?

The interrelationship between POD24 and these prognostic assessment tools.

The pathogenesis of early disease progression.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00277-024-06079-y.

Abstract

Indolent B-cell non-Hodgkin lymphomas(B-NHL) encompass a heterogeneous category of lymphomas characterized by a wide range of pathological subtypes. With the application of chemoimmunotherapy with rituximab (R-chemo), the prognosis of patients has improved considerably, with a 10-year survival rate of 60-80%. Despite these advancements, a significant number of patients still experience disease progression during or shortly after initial treatment. Those who progress within the first 24 months (POD24) continue to face a notably worse prognosis. This study aims to explore the significance of POD24 in predicting the prognosis of different subtypes of indolent B-cell NHL through a comprehensive literature review. The investigation extends to examining the existing prognostic assessment tools and evaluating the interrelationship between POD24 and these tools. By synthesizing relevant research findings, this study seeks to contribute to the current understanding of the role POD24 plays in prognostic evaluation and its potential implications in guiding clinical decision-making for patients with indolent B-cell NHL.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00277-024-06079-y.

Introduction

Indolent B-cell non-Hodgkin lymphomas(B-NHL) is a group of lymphomas with diverse pathologic types, accounting for approximately 35–45% of non-Hodgkin lymphomas, the majority of which progress slowly but will not be cured. With the use of anti-CD20 monoclonal antibodies, the 10-year overall survival (OS) rate of these patients has increased to about 60–80%, with some variability observed among different subtypes [1–3]. The choice of treatment for indolent lymphoma is dependent on the patient’s basic condition and disease stage. Patients with a low tumor burden and no treatment indicators may undergo watch-and-wait or receive rituximab as monotherapy based on individual patient characteristics. Chemoimmunotherapy (CIT) regimens with CD20 monoclonal antibody in combination with chemotherapy have become the current first-line treatment option [4–10]. Given the range of treatment options available, accurately assessing prognosis is essential for selecting the most appropriate therapeutic strategy. The association between POD within 24 months (POD24) and poor prognosis was proposed for the first time in follicular lymphoma (FL) in 2015, after which the prognostic value of POD24 in indolent lymphoma was gradually recognized [11]. The aim of this paper is to review the significance of POD24 in the prognosis of various types of indolent B-NHL: including follicular lymphoma, marginal zone lymphoma, lymphoplasmacytic lymphoma/Waldenström macroglobulinemia, and chronic lymphocytic leukemia/small lymphocytic lymphoma, the relationship between POD24 and currently used clinical prognostic tools, as well as the limitations of POD24 and future research directions through literature study.

Definition of early POD

Existing research on indolent lymphoma predominantly utilized a modified definition of POD24, which means disease progression or relapse within 24 months from the initiation of first-line therapy. The 24-month time point was chosen since approximately 20% of FL patients always experience POD within 2 years after receiving first-line therapy [9, 12, 13]. In subsequent studies, researchers have further validated that the peak of disease progression is within 24 months of diagnosis [11].Consequently, 24 months is now be widely used as the time node for early relapse in majority of studies [14–21]. There are also different views on the starting point of POD24; in the study by Casulo et al. the starting point of POD24 was defined as the time of disease diagnosis, and the R-CHOP(rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) was administrated as the first-line treatment regimen [11]. And in another study that specifically explored predictive models for early POD [18] researchers discovered that when the POD24 starting point was defined as from the onset of treatment, the number of patients in the POD24 group increased by eight. Although this number is small, the prognosis for these individuals was found to be dismal. The median survival time for these patients was 3.1 years (P < 0.0001), with only one patient still alive after 7.6 years. This implies that defining the POD24 starting point as the diagnosis of disease excludes a proportion of patients with a poorer prognosis, and therefore most subsequent studies have defined the POD24 starting point as onset of initial treatment [18–20, 22].Several scholars, including Maurer et al., have established a uniform interpretation of POD to encompass relapse, disease progression necessitating unplanned re-treatment post initial management, and fatalities from any cause. This description aligns closely with the concept of Event-Free Survival (EFS) [3, 16, 23–26]. In contrast, Sortais et al. [14]. offer a more streamlined definition focusing solely on disease progression or lymphoma-related deaths. Another study by Freeman et al. [15] expands the POD definition to include disease progression, relapse, or fatalities resulting from lymphoma or treatment-related complications, a definition that holds practical merit. Freeman et al.‘s delineation of early POD emerges as the most cogent and comprehensive among the various interpretations. Their definition encapsulates not only disease progression and relapse but also accounts for fatalities attributed to lymphoma or treatment toxicity. By encompassing these diverse outcomes, Freeman et al.’s definition offers a nuanced understanding of early POD that considers the multifaceted nature of disease progression in indolent B-cell non-Hodgkin lymphomas. Additionally, some researchers have found that the earlier the POD, the higher the risk of death [27]. And for patients receiving rituximab-based immunochemotherapy regimens, the POD60(POD occurs within 60 months) after the initiation of therapy) is still associated with poor outcome [22]. Therefore, the early POD time points could be further investigated.

Impact of POD24 on the prognosis of distinct types of indolent B-NHL

Impact of POD24 on the prognosis of FL patients

FL is considered a typical indolent B-NHL of germinal center (GC) origin, accounting for 20-25% of all non-Hodgkin lymphoma, approximately 85% of patients have chromosomal translocation t (14;18) (q32;q21), which causes an overexpression of BCL-2 protein, resulting in an excessive proliferation of lymphocytes [28, 29]. In the modern era with front-line chemoimmunotherapy, substantial advances have been made in the OS of patients with FL, with a 10-year OS rate of approximately 80% [25]. Despite the improved outcome for majority of patients, approximately 20% of patients still experience progression of disease within two years from initial treatment [9, 12, 15]. And the negative prognostic impact of early disease progression has been validated in different treatment regimens. Casulo et al. [11] first reported the association between POD24 and poor prognosis according to the analysis on the data of 588 FL patients treated with R-CHOP. Patients in this study were divided into two groups: early POD (POD within 24 months from diagnosis, 19%) and the reference group (without POD within 24 months of diagnosis, 71%).The 5-year OS for early POD patients was only 50%, but the reference group had a 5-year OS of 90%. This trend was maintained even when we adjusted through FL International Prognostic Index (FLIPI), with a hazard ratio (HR) of 6.44 (95% confidence interval [CI], 4.33–9.58). And it was verified by the validation cohort consisting of 147 FL patients in stages II to IV who received R-CHOP as a first-line therapy. Subsequent studies have further confirmed the strong association between POD24 and poor prognosis in patients treated with various rituximab-based regimens, including R-CHOP [17, 20, 23, 30], BR (Bendamustine and Rituximab), and R-CVP (rituximab, cyclophosphamide, vincristine, and prednisone) [11, 18, 23]. In the early 21st century, research demonstrated that R-CHOP significantly improved outcomes compared to CHOP alone, establishing it as the preferred first-line treatment for FL [31]. However, the search for more effective treatments continued, leading to the development of the BR regimen, which offered fewer toxicities and a substantial improvement in progression-free survival (PFS) compared to R-CHOP and R-CVP, though it did not extend OS. The pivotal StiL (Study Group Indolent Lymphomas) trial provided strong evidence supporting BR’s superiority over R-CHOP. This phase III study included 514 patients with previously untreated indolent and mantle cell lymphomas and found that BR significantly prolonged PFS, with a median of 69.5 months compared to 31.2 months for R-CHOP. BR also resulted in fewer toxicities, particularly hematological side effects such as neutropenia, and less alopecia and peripheral neuropathy [9]. Flinn’s study further confirmed these findings, showing a five-year PFS rate of 65.5% for BR versus 55.8% for R-CHOP/R-CVP, with a significant hazard ratio for PFS of 0.61 (95% CI, 0.45–0.85; P = 0.0025). The BR regimen also demonstrated advantages in event-free survival and duration of response, though no significant difference in OS was observed. The safety profiles of BR, R-CHOP, and R-CVP were consistent with expectations, and no new safety concerns emerged during long-term follow-up [32]. Mondello’s study echoed these results, reporting a median PFS of 15 years for BR compared to 11.7 years for R-CHOP (P = 0.03) [33]. As a result, BR has become the preferred first-line treatment in the United States [34]. POD24 was also associated with a significantly inferior outcome even though FL patients treated with frontline BR. In Freeman et al. study, the subset of patients who experienced POD24 exhibited a notably lower two-year OS rate of 38%, which was significantly inferior than the two-year OS rate of 92% observed in the total patient population treated with the BR regimen. The present study comprised a cohort of 296 individuals diagnosed with symptomatic advanced FL, classified histologically as grades 1–3 A. Among these patients, 13% experienced POD24. Notably, the implementation of BR treatment showed a significant reduction in the occurrence of POD24 complications when compared to patients who received R-CVP treatment (13% vs. 23%, P = 0.001) [15]. In another large trial cohort consisting of 984 FL patients, POD24 was associated with 17-fold increased mortality compared to no progression within 24 months for patients received BR treatment(n = 150). There was no statistically significant difference in the incidence of POD24 between R-CHOP and BR patients (50.9% vs. 60%) [22]. For patients treated with R-CVP, the predictive role of POD24 still applies [18, 23] .

Obinutuzumab-based regimens have been shown to reduce the risk of early disease progression in patients with FL compared to rituximab-based immunochemotherapy. In the GALLIUM randomized controlled trial, 1,202 patients were randomly assigned to receive either Obinutuzumab- or rituximab-based chemotherapy. After a median follow-up of 34.5 months, interim results demonstrated that Obinutuzumab significantly lowered the risk of progression, relapse, or death. The 3-year progression-free survival rates were 80.0% in the Obinutuzumab group and 73.3% in the rituximab group (hazard ratio 0.66, 95% CI: 0.51–0.85; P = 0.001) [27, 35]. Further analysis by Seymour et al. (excluding 131 patients lost to follow-up) revealed that patients who experienced POD24 had a significantly higher risk of mortality compared to those without progression. The hazard ratio for overall survival in this group was 12.2 (95% CI: 5.6–26.5), even after adjusting for age. The cumulative incidence of POD24 was 10.1% (95% CI: 8–12%) in the Obinutuzumab group and 17.4% (95% CI: 14–20%) in the rituximab group, reflecting a 46.0% reduction in risk of early progression with Obinutuzumab. For patients with POD24, post-progression survival was similar in both treatment arms, with a median follow-up of 22.6 months. However, across the entire cohort, patients who progressed within the first 24 months faced a higher risk of mortality [27].This indicates that the impact of POD24 on prognosis appears to hold consistent value across different immunochemotherapy protocols for FL, and particularly in relation to the R-CHOP regimen.

In symptomatic patients with low volume tumor mass, and in case of slow disease progression, therapy with rituximab alone may be considered [36]. The prognostic effect of POD24 for patients receiving R monotherapy is a subject of controversy. In some study, we can find that POD24 did not significantly influence the five-year survival outcomes for patients receiving rituximab alone [14, 24]. Several other studies have shown that although POD24 is still being an independent risk factor for poor prognosis, the role of POD24 in influencing survival is not as significant when compared with patients receiving immunochemotherapy [22, 37, 38]. However, the results were reversed in the study by Casulo et al. [17]. The study examined the relationship between POD24 and patient prognosis depending on the treatment regimen. Specifically, the study compared the association between POD24 and prognosis between patients who received chemotherapy plus R (chemotherapy plus rituximab) or chemotherapy only and those who received R monotherapy. For these three groups of patients, the HR was 5.39 (n = 2252), 4.26 (n = 2335), and 6.57 (n = 265), respectively. This means that the prognostic impact of POD24 was more significant for patients who received only R monotherapy.

Lenalidomide and rituximab (R2) have been utilized as a combination regimen in the treatment of FL with the primary objective of enhancing therapeutic efficacy while mitigating drug toxicities. Clinical trials have shown that the R2 regimen has similar or even superior efficacy compared to immunochemotherapy in terms of complete response (CR) rates and PFS rates. Additionally, the R2 regimen has been associated with a lower incidence of grade 3–4 adverse effects. Due to its favorable efficacy and safety profile, the 2024 NCCN (National Comprehensive Cancer Network) guidelines had recommend R2 as a first-line treatment option for FL [36, 39, 40]. In a cohort study involving 74 patients treated with the R2 regimen, notable disparities were observed in the 5-year and 10-year OS rates between those who developed POD24 and those who did not. Patients experiencing POD24 exhibited a 5-year OS rate of 69% and a 10-year OS rate of 59%, while patients without POD24 demonstrated a significantly higher 5-year OS rate of 92% and a 10-year OS rate of 77%.(P < 0.0001) [38]. For other chemotherapy-free treatment, including rituximab plus galiximab regimen and rituximab plus epratuzumab regimen, this trend was maintained [41].

Therefore, it is reasonable to believe that POD24 can serve as a good predictor of poor prognosis for FL patients receiving first-line therapy, regardless of the treatment regimen (Table 1). The presence of poor prognosis in FL patients following early POD may potentially be attributed to a higher occurrence of histological transformation (HT). Studies have demonstrated that regardless of the chemotherapy regimen used—whether R-CHOP, BR, or G-chemotherapy—the rate of HT in patients with early progression is notably high, ranging from 19% to 70%. This transformation usually occurs within the first 24 months after diagnosis. For those who experience transformation, survival outcomes are generally unfavorable, irrespective of their POD24 status [15, 20, 42, 43]. In a study conducted by Maurer et al., which included two patient cohorts, POD24 patients who experienced transformation had significantly lower 5-year OS rates compared to POD24 patients with FL progression. In cohort 1, the 5-year OS rates were 27% versus 54% (HR 0.36, 95% CI 0.19–0.70), while in cohort 2, the rates were 31% versus 61% (HR 0.45, 95% CI 0.22–0.89) [44].In Muntañola’s study, among 188 patients with progression, only FLIPI (HR 3.57; p < 0.001) and HT (HR 3.41; p < 0.001) were considered independent prognostic factors, while POD24 failed to predict survival after relapse (p = 0.85) [45].

Table 1.

Differences in prognosis between patients with early versus late disease progression under different treatment regimens for FL patients

Study	Number of patients	Definition of POD24	Treatment	POD24 rate	Pod24 OS	non-POD24 OS	HR/P
Casulo 11	N=588	POD24: POD occurred within 24 months after diagnosis	R-CHOP	19%	50% (5years)	90% (5years)	HR = 19.8
Maurer23	MER(N = 920)	EFS12/24: EFS24 was defined as EFS status 24 months from diagnosis.	WW(n = 326), R -mon(n = 111); immunochemotherapy (n = 349); Other therapies(n = 134)	17% (EFS12)	SMR = 13.2 (no EFS12:17.63)	SMR = 0.37 (EFS12:0.74)	-
	Lyon(N = 412)		WW(n = 82), R -mon(n = 43); immunochemotherapy (n = 243); Other therapies(n = 44)	18% (EFS12)	SMR = 7.22 (EFS12:19.1)	SMR = 0.90 (EFS12:1.03)	-
Jurinovic18	GLSG (n = 151)	POD24: POD occurred within 24 months after first-line treatment initiation	R-CHOP + IFN Maintenance	17%	41% (5years)	91% (5years)	HR = 9.72
	BCCA (n = 107)		R-CVP + R maintenance	13%	26% (5years)	86% (5years)	HR = 11.93
Provencio24	1074	EFS12/EFS24: EFS12 /24indicates event-free survival at 12 months or 24 months	CT with anthracyclines(n = 155 ) CT without anthracyclines (n = 96 ) With anthracyclines and rituximab (n = 610) Without anthracyclines and rituximab (n = 109 ) Rituximab monotherapy (n = 34) Surgery (n = 13) Observation(n = 18)	-	SMR = 8.42 (no EFS12:10.27)	SMR = 1.3 (EFS12:1.75)	-
Bachy16	1135	EFS24: EFS24 was defined as EFS status 24 months from diagnosis	R-CHOP (n = 840); R-CVP(253); R-FCM(n = 42)	25%	63% (5years)	92% (5years)	P < 0.0001,
Freeman15	N = 296	POD24: POD occurred within 24 months after initiation of systemic therapy	BR	13%	38% (2 years)	92% (2years)	-
Sortais14	N = 317	POD24:POD or death from POD occurred within 24 months after diagnosis	R-based(n = 259); radiotherapy alone (n = 9); anthracycline based regimens without R(n = 15); GR(n = 3); other regimens (n = 7)	18.9%	82 (5years)	93.3% (5years)
Yoon20	N = 324	POD24: POD occurred within 24 months after treatment,	R-CVP (n = 152),	16.4%	-	-	P = 0.192
			R-CHOP (n = 111),	25.2%
			BR ( n = 61)	8.2%
Moccia38	N = 318	POD24: POD occurred within 24 months after chemoimmunotherapy	R monotherapy(n = 244) R2(n = 74)	27%	69% (5years)	92% (5years)	P < 0·0001
					59% (10 years)	77% (10years)
Casulo17	N = 5225	-	Chemo(n = 2545); R-Chemo(n = 2410); R monotherapy(n = 270)	29.3%	71.2% (5years)	93.6% (5years)	-
Weibull22	N = 948	POD12: POD occurred within 12 months after first-line treatment initiation	R-chemo (R-CHOP (308), BR (105), R-CVP, R-FC (n = 519); R monotherapy (n = 156); other treatments (n = 156)	28%	34% (5years)	78% (5years)	-
		P0D24 : POD occurred within 24 months after first-line treatment initiation			46% (5years)	82% (5years)	-
		POD60: POD occurred within 60 months after first-line treatment initiation			57% (5years)	83% (5years)	-

FL: follicular lymphoma; POD = Progression of disease; event-free survival (EFS) = defined as time from diagnosis until relapse or progression, unplanned retreatment of lymphoma after initial management; The standardized mortality ratio (SMR) = the ratio of observed to expected deaths, based on the age and sex-matched general population; HR = hazard ratio; R-CHOP = rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone; IFN = interfern ; R-CVP = rituximab, cyclophosphamide, vincristine, and prednisone; RFCM = rituximab plus fludarabine, cyclophosphamide and mitoxantrone; BR = bendamustine, Rituximab; R2 = Lenalidomide rituximab; GR = Obinutuzumab Lenalidomide; Chemo = chemotherapy ; CT = chemotherapy

Impact of POD24 on the prognosis of MZL patients

Marginal zone lymphoma (MZL) is a malignant proliferative disease that arises from the marginal zone B-cells in the lymphoid tissue, including three subtypes: extranodal mucosa-associated lymphoid tissue type (MALT lymphoma), splenic MZL (SMZL) and nodal MZL (NMZL). MZL is the third most common type of B-cell lymphoma, which usually exhibits an indolent course and has a median survival of 10 years [46, 47]. To clarify whether POD24 is similarly predictive of prognosis in MZL, Luminari et al. [48] analyzed data on 321 patients with treatment-naïve MZL(including various subtypes) in the NF10 trial group and found that the incidence of POD24 was 18% (59 out of 321). The 3-year OS for this subset of patients with early disease progression was 53%(95% CI 37–67%), and the 3-year OS for patients in the non-POD24 group was 88%(95% CI 89–98%) with an HR of 19.5(95% CI 8.4–45.4). Another investigation was conducted on 401 patients with MALT lymphoma who were treated with nitrogen mustard phenylbutyrate, R, or a combination of the two drugs. It was found that the incidence of POD24 in these patients was 17%. For these patients who progressed early, the 5- and 10-year overall survival rates were 80% (95% CI: 69–88%) and 64% (95% CI: 45–78%), respectively. While for those patients without POD24, their 5- and 10-year overall survival rates were 91% (95% CI: 87–94%) and 85% (95% CI: 79–90%), respectively. The POD24 group had an HR of 2.42 (95% CI: 1.35–4.35) compared to the non-POD24 group, and this result was confirmed in a validation cohort of 224 patients with MALT lymphoma, suggesting that POD24 is strongly associated with OS in patients with MALT [49]. Moreover, in a retrospective analysis upon 106 patients with SMZL who had treatment indicators and received regular treatment, Lu et al. [50]discovered that the independent prognostic significance of POD24 is supported by both single-factor and multi-factor prognostic analyses. Detailed data for all studies above are shown in Table 2. It’s worth noting that the association between POD24 and prognosis appeared to be weaker for patients who did not receive treatment [48]. So when a patient receives R monotherapy, does POD24 still have an influential role in patient prognosis? A retrospective multicenter study confirmed that, whether the MZL patients received first-line treatment with R monotherapy or CIT, with a 5-year OS of 75% in the POD24 group compared to 88% in the non-POD24 group for patients treated with R monotherapy (P = 0.07) [51]. As demonstrated above, early POD in patients with MALT and SMZL is indicative of a poor prognosis. However, relevant research evidence for patients with NMZL subtype is temporarily lacking.

Table 2.

Differences in prognosis between patients with early versus late disease progression in patients with different subtypes of MZL

Study	N	Classification	Definition of POD24	Treatment	POD24 rate	OS(POD24)	OS(non-POD24)	HR/P
Luminar48	N = 321	MZL	POD24; POD occurred within 24 months after diagnosis	Alk-Mono (n = 16) ;R-Mono(n = 30 );R-Alkylating(n = 83);R-CHOP(n = 48); BR (n = 112) ;R-Fludarabinen(= 3 ) ;Other(n = 21)	18%	53% (3years)	88% (3years)	HR = 19.5 p < 0.001
Conconi49	N = 401	MALT	POD2;progression within 24 months from the start of first-line treatment	Chemotherapy only(n = 131);Rituximab and chemotherapy(n = 132); R-Mono  (n = 138)	18%	64% (10years)	85% (10years)	HR = 2.42
Lu50	N = 106	SMZL	POD24: POD occurred within 24 from diagnosis	CHOP/COP/FC(n = 20); R + CHOP /COP/FC(n = 54); splenectomy (n = 11); R only(n = 4); combination of the proposed approaches (n = 17)	16.0%	-	-	HR = 20.116
Epperla51	N = 364	MZL	POD24: POD occurred within 24 from systemic therapy initiation	R-Mono	28%	75% (5years)	88% (5years)	p = 0.007
				ICT		69% (5years)	94% (5years)	p < 0.0001

MZL: Marginal zone lymphomas; POD = Progression of disease; HR = Hazard Rates; Alk = alkylating agent; B = Bendamustine ; R-Mono = rituximab monotherapy; CHOP = cyclophosphamide, doxorubicin, vincristine, and prednisone; ICT = immunochemotherapy; FC = fludarabine, cyclophosphamide

Impact of POD24 on the prognosis of patients with LPL/WM

Lymphoplasmacytic lymphoma (LPL) is a rare hematologic tumor characterized by the proliferation of mature small B lymphocytes, plasmacytoid lymphocytes, and plasma cells. Waldenström macroglobulinemia(WM) is a specific subtype of LPL. It is diagnosed when there is the presence of an IgM monoclonal gammopathy [52, 53]. LPL/WM is generally categorized as an indolent lymphoma, survival and prognosis among patients exhibits considerable variability, typically spanning a range of 5 to 10 years on average [53, 54]. Guidez et al. [55] found that the onset of progression and the initiation of second-line treatment, but not the response, determined the survival of WM patients after first-line treatment. However, in a study involving 472 patients diagnosed with symptomatic WM, an investigation into the impact of POD24 on prognosis yielded noteworthy findings. The researchers observed that patients in the POD24 group exhibited a 3-year OS rate of 75%, while the non-POD24 group demonstrated a higher 3-year OS rate of 93%. However, the evaluation of POD24’s impact on prognosis in symptomatic WM was complex, as deviations from the proportional hazard assumption hindered accurate assessment [56]. Additionally, the study revealed that the timing of disease progression did not significantly affect survival rates. Patients whose disease progressed within 24 months displayed similar survival outcomes compared to those whose disease progressed after this timeframe. Notably, patients whose disease progressed after two years showed a 3-year OS rate of 84%. Furthermore, the study indicated that the non-POD24 group, when considering patients with symptomatic WM, had a median EFS of 43 months after the initial two-year period. These findings highlight the multifactorial nature of prognosis evaluation in LPL/WM and underscore the need for further exploration into the underlying mechanisms influencing patient outcomes.

The impact of POD24 on the prognosis of CLL/SLL patients

Chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL) is a hematological tumor originating from mature B cells. CLL and SLL are different manifestations of the same disease. Diagnosis of CLL relies on the detection of a specific threshold of lymphocytes within the bone marrow and peripheral blood. Conversely, SLL manifests as a condition where lymphocytes form tumors primarily in lymph nodes or other lymphoid tissues. Notably, individuals with SLL typically exhibit a normal count of lymphocytes in both the bone marrow and peripheral blood, differentiating it from CLL [57, 58]. CLL/SLL is classified as an indolent neoplasm, exhibiting a 5-year OS rate of roughly 90% and a 10-year OS rate of approximately 80% [59]. To validate the predictive role of early disease progression for poor prognosis in patients with CLL, Ahn categorized 829 patients with CLL into 3 groups: early POD (progression < 2 years after treatment initiation), late POD (progression ≥ 2 years after treatment initiation), and no POD that did not experience disease progression during the follow-up period. The 3-year OS of these patients were 65.9%, 94.4%, and 85.5%, respectively [60]. The presence of early POD demonstrated a strong association with inferior OS outcomes across the entire patient cohort (HR 3.6, P < 0.01). This association remained significant in patients who received specific treatment regimens, including fludarabine, cyclophosphamide plus rituximab, and bendamustine plus rituximab(P < 0.05). Notably, early POD occurring within two years following initial therapy emerged as a robust clinical prognostic factor independently predicting inferior OS in patients diagnosed with CLL. An additional study revealed a significant correlation between the duration from diagnosis to second treatment (TT2T) and an unfavorable prognosis in patients diagnosed with CLL. This finding suggests the potential for TT2T to supplant OS as the primary endpoint in future clinical trials [61]. Among these patients, only 14-25.5% received targeted therapy. In a phase 3 open-label trial, Venetoclax combined with Obinutuzumab (with or without Ibrutinib) significantly increased PFS from 75.5 to 90.5% [62]. Similarly, another phase 3 trial demonstrated improved progression-free survival with Ibrutinib-rituximab [63]. However, there are currently no studies specifically examining the prognostic impact of POD24 in CLL or SLL patients undergoing targeted therapies, highlighting the need for further research. In a study specifically focusing on elderly patients, Ammann et al. observed that individuals with a treatment interval exceeding 3 years between first- and second-line treatments exhibited a marginal enhancement in 5-year relative survival (P < 0.01) compared to those with treatment intervals ranging from 6 months-2 years and 2–3 years. However, despite this slight improvement, the 5-year survival rate for these patients remained considerably lower when compared to the overall population [64].

POD24 and Other Clinical risk stratification systems and prognostic scores

Clinical prediction model of FL

FLIPI

The FLIPI scoring system stratifies patients into three risk groups based on the number of adverse factors (detailed in Supplement Table 1). A relative risk ratio of 4.3 highlights the significant distinction between high- and low-risk groups, validating FLIPI as an effective prognostic tool [65]. Even in the rituximab era, FLIPI has been validated and remains relevant [66–68]. Sevilla et al. found that FLIPI outperformed other prognostic tools like FLIPI-2, PRIMA-PI, FLEX, and m7-FLIPI, confirming its accuracy in identifying varying prognostic profiles [69]. This finding suggests that even in the era of rituximab, FLIPI remains a widely utilized and highly effective prognostic assessment tool [70].

FLIPI-2

With rituximab’s introduction and β2-microglobulin (β2-MG) identified as a poor prognostic factor, Federico et al. [71] conducted a prospective study with 1,000 FL patients to develop a new prognostic model. Five factors, including β2-MG and nodal size, were linked to PFS, categorizing patients into low-, intermediate-, and high-risk groups with 3-year PFS rates of 91%, 69%, and 51% (P < 0.00001).While some studies suggest FLIPI-2 may be more effective than FLIPI [72, 73], conflicting evidence calls for further validation before FLIPI-2 can replace FLIPI as the primary prognostic tool [69]. As a result, FLIPI-2 has not yet replaced FLIPI as the predominant scoring system in clinical and experimental settings due to the need for additional validation studies [16, 72, 73].

PRIMA-PI

The PRIMA-PI is a simplified prognostic tool that classifies patients into three risk groups based on β2MG levels and bone marrow involvement. The 5-year PFS rates for low-, intermediate-, and high-risk groups were 69%, 55%, and 37%, respectively. PRIMA-PI is easier to implement than FLIPI and FLIPI2, avoiding errors in lymph node counts [16]. Additionally, Kimby et al. found that for 291 FL patients treated with rituximab monotherapy or rituximab plus interferon (R plus IFN), PRIMA-PI outperformed FLIPI in risk stratification, and FLIPI was ineffective in predicting time-to-treatment failure (TTF) [74].

FLEX

FL Evaluation Index (FLEX) is the pioneering prognostic model specifically developed for patients undergoing a treatment regimen involving the combination of Obinutuzumab and bendamustine. In a comprehensive analysis involving two distinct cohorts, namely the phase 3 GALLIUM trial group [35]comprising 1202 patients and the SABRINA validation group consisting of 1004 individuals with FL, the FLEX showed notable advantages in terms of discerning inter-group variances in 2- and 3-year PFS as compared to FLIPI, FLIPI-2, and PRIMA-PI. However, it is worth noting that FLIPI demonstrated a stronger predictive ability for OS in the trial group [75].

Genetic prediction models of FL

m7-FLIPI

Positive and negative predictive values of the m7-FLIPI model for 5-year PFS rates were 64% and 78%, respectively, and in a validation cohort of 107 FL patients from the British Columbia Cancer Agency (BCCA), they were 72% and 68%, respectively. Since the experimental group and validation group in this trial received R-CHOP and R-CVP, respectively, the prediction value of m7-FLIPI for other treatment regimens (such as bendamustine or alternative anti-CD20 antibody treatment regimens) requires further validation [76]. Notably, approximately half of the patients who were classified as high-risk by FLIPI were classified as low-risk by M7-FLIPI in these two cohorts, and the prognosis of these patients did not differ from those classified as low-risk by FLIPI.

23-gene scoring

Huet et al. [77] constructed a 23-gene predictive model associated with the prognosis of FL utilizing biopsy tissue information from 134 untreated FL patients with high tumor burden in the PRIMA phase III trial. The model was validated in three independent validation cohorts and can predict the risk of PFS at diagnosis. However, in their investigation, the proportion of patients at high risk of progression varied slightly between validation cohorts, and the study did not include patients treated with BR.

Clinical prediction model of other indolent B-cell lymphomas

Prediction model of MZL

Currently, there are no universally accepted prognostic indices for MALT. The MALT-IPI, developed from the IELSG-19 trial, identifies advanced-stage disease, age ≥ 70, and elevated serum LDH as key factors ( detailed in Supplement Table 2). It stratifies patients into 3 risk groups with 5-year EFS rates of 70%, 56%, and 29%. Validation in a cohort of 633 patients from three independent studies confirmed its accuracy [78]. In another study from the Mayo Molecular Epidemiology Resource, MALT-IPI outperformed the International Prognostic Index (IPI) and FLIPI in predicting PFS and OS. The concordance statistics for EFS were 0.593 for IPI, 0.598 for FLIPI, and 0.612 for MALT-IPI. For OS, the c-statistics were 0.689 for IPI, 0.683 for FLIPI, and 0.714 for MALT-IPI [26]. A revised MALT-IPI for EMZL improved risk identification, increasing high-risk cases from 17–26% [79]. Additionally, a new prognostic index, known as MZL-IPI, based on five clinical variables (LDH, hemoglobin, platelets, lymphocytes, and MZL subtype) further refined patient stratification, showing superior predictive accuracy compared to IPI, FLIPI, and MALT-IPI, with 5-year PFS rates of 85%, 66%, and 37% for low-, intermediate-, and high-risk groups, respectively [80].

Prediction model of LPL/WM

International prognostic scoring system for Waldenström macroglobulinemia (IPSS-WM) was developed using data from 587 patients and classified them into three risk groups: low, intermediate, and high, representing 27%, 38%, and 35% of the cohort, respectively. The corresponding 5-year survival rates for these groups were 87%, 68%, and 36% (P < 0.001) [81]. Since its introduction, IPSS-WM has served as a key tool for risk stratification in treatment-naive patients. Although widely used, IPSS-WM’s validity needs reassessment due to new mutations and therapies introduced since 2001, and it excludes factors like non-WM-related deaths and LDH. To address these limitations, the revised IPSS-WM (rIPSSWM) was introduced, incorporating age, β2-MG, serum albumin, and LDH to stratify patients into five risk groups. The 3-year WM-related mortality rates ranged from 0 to 48%, while 10-year survival rates varied from 84–9% [82]. However, rIPSSWM does not account for the MYD88 L265P mutation, a key prognostic factor [83]. A newer model, the Modified Staging System for WM (MSS-WM), based on age, LDH, and serum albumin, demonstrated 5-year OS rates of 93%, 82%, 69%, and 55% for low- to high-risk groups, respectively, and was validated in a separate cohort [84] .

Prediction model of CLL/SLL

The CLL International Prognostic Index (CLL-IPI) is a widely used scoring system based on five key prognostic factors: TP53 abnormalities, IGHV mutation status, serum β2-microglobulin levels, clinical stage, and age. It stratifies CLL patients into four risk groups (low, intermediate, high, and very high), with 5-year overall survival (OS) rates of 93%, 79%, 63%, and 23%, respectively [85]. The CLL-IPI retains its prognostic value for predicting PFS with targeted therapies, but its effectiveness in predicting overall survival appears to be reduced [86]. As targeted therapies have improved survival, new prognostic models Four-Factor Prognostic Model have emerged. One model for patients treated with ibrutinib uses TP53 abnormalities, prior treatments, serum β2-microglobulin, and LDH to categorize patients into low-, intermediate-, and high-risk groups, with 3-year PFS rates of 87%, 74%, and 47%, and OS rates of 93%, 83%, and 63% [87]. This model has demonstrated significance in both treatment-naive and relapsed/refractory CLL patient populations. For high-risk patients with poor outcomes on existing therapies (such as ibrutinib, idelalisib and venetoclax), a derived model incorporates serum β2-microglobulin, LDH, hemoglobin, and treatment timing to stratify patients into low-, intermediate-, and high-risk groups. This model has showed prognostic value in the internal validation dataset (C-statistic = 0.79, 95% CI 0.56–0.97, log-rank P = 0.0003), and was validated across three external validation cohorts [88].

Relationship between POD24 and predictive scoring systems

In order to better understand the characteristics of patients with indolent lymphomas and improve treatment outcomes, researchers have been focusing on exploring the risk factors closely associated with POD24. Extensive studies on FL have developed various predictive models for POD24. While the FLIPI is commonly used in clinical practice, its ability to predict POD24 is limited. The sensitivity of FLIPI for predicting POD24 ranges from 53 to 78%, with specificity around 60%, leading to concerns about overestimating the likelihood of early progression [18, 75]. Similarly, the PRIMA-PI scoring system has demonstrated varying sensitivity (26-65%) and specificity (48-88%) across studies, with limited effectiveness in predicting POD24. Although combining PRIMA-PI with the marker KI-67 improves sensitivity to 92%, specificity remains low at 48%, with an area under the curve (AUC) of 0.70 [89, 90]. The 23-gene prediction model showed that high-risk patients were twice as likely to develop POD24 compared to low-risk patients (38% vs. 19%), but its sensitivity and specificity were only 43% and 79%, respectively [77]. Notably, this model included only patients with high tumor burden, and its applicability to those with low tumor burden still requires validation. The FLEX scoring system, utilizing more contemporary treatment regimens, offers improved specificity in predicting POD24 compared to FLIPI, FLIPI-2, and PRIMA-PI. However, its sensitivity and specificity are still moderate at 60% and 68%, respectively [75]. Several POD24-specific predictive tools have been developed, such as POD24-PI, total lesion glycolysis (TLG), and FLIPI24, as well as the use of metrics like not achieving complete remission (non-CR) or complete metabolic remission (non-CMR). POD24-PI, for example, integrates gene mutations (EZH2, FOXO1, and EP300), clinical risk factors (FLIPI), and performance status (ECOG PS) to create a composite score. This score is calculated by summing the weighted predictive values derived from Lasso coefficients. A score above 0.71 categorizes patients into a high-risk group, while a score below 0.71 indicates low risk. In two independent cohorts, GLSG [91] and BCCA [76], POD24-PI demonstrated reasonable predictive performance, with sensitivities of 78% and 67% and specificities of 61% and 73%, respectively. While it showed higher sensitivity compared to m7-FLIPI and greater specificity than FLIPI, m7-FLIPI remains the most accurate for predicting POD24, with sensitivity and specificity values of 76% and 77%, respectively [18]. The FLIPI24 scoring system has also shown promise, categorizing patients into various risk groups (very low to very high) and visualizing these categories through Kaplan-Meier curves. It has outperformed traditional models such as FLIPI and PRIMA-PI in predicting EFS and OS in both internal and external validation cohorts [92]. Despite advancements, existing tools often fall short of delivering both high sensitivity and specificity for POD24 prediction. For example, a predictive model based on initial treatment response (non-CR/CMR) and high TLG exhibited high specificity for predicting POD24 ((100%, Table 3) but had low sensitivity (56%) and was limited by a small sample size of just 45 patients [19]. In addition, emerging research suggests that immune infiltration within tumors may play a key role in POD24 development. Specifically, follicular lymphoma patients with low expression of PD-L2 appear to be more susceptible to experiencing POD24 [93]. Additionally, recent studies have linked POD24 with histological transformation [15, 20, 30, 42, 49]. A separate study has provided evidence of a significant correlation between del(17p) and the heightened risk of early POD [60] .Moreover, therapeutic interventions can be perceived as a selective pressure, promoting the survival and growth advantage of clones with pre-existing somatic mutations, such as TP53 [94]. Additionally, treatment itself may facilitate the occurrence of new genomic alterations, contributing to an increase in clonal complexity. These gene mutations associated with treatment have been consistently associated with unfavorable prognoses [95]. Given that in addition to pathogenic single nucleotide variations and abnormal gene expression, copy number aberration can also affect prognosis, Gao et al. [96] investigated the pathogenesis of POD24 from these three aspects. The study found that the mutation frequency of the HIST1H1D gene involved in the formation of nucleosome structure was significantly higher in the POD24 subgroup than in the non-POD24 subgroup (17% vs. 4%, P < 0.05). Gene-level copy number increase analysis discovered that gains of 6p22.2 (HIST1H1D; P = 0.02; HR 4.75) and 18q21.33 (BCL2; P = 0.03; HR 3.23), loss of 1p36.13 (NBPF1; P < 0.01; HR 5.73) could predict POD24 independently of FLIPI. The transcriptomes of 41 FL samples were characterized. A unique gene expression profile was observed in the POD24 subtype compared to the non-POD24 subtype, including a set of up-regulated genes (n = 434), such as MME, CXCR1, CD5L, HLA-G, IL27, ZNF365, and several genes associated with major histocompatibility complex (MHC) class I molecules, as well as a set of down-regulated genes (n = 840), such as CCND1, CXCL8, TP73, and histone family member genes. In this regard, the occurrence of early POD may assume a potential role as a post-treatment risk stratification tool, offering a straightforward assessment of clonal fitness following first-line treatment, dispensing the requirement for further complex or costly genomic analyses. In a discovery cohort of 97 tumor biopsies and 42 matched blood samples from 44 patients diagnosed with FL grades 1–3 A, 22 with transformation (tFL) and 22 with relapse without transformation (nFL). Through multi-omics analysis of longitudinal FL samples, the study identified key genomic alterations that occurred early in the disease. Mutations in CREBBP and KMT2D were among the earliest events, persisting throughout disease progression and transformation into aggressive lymphoma. Early and stable copy number alterations, including gains in 7p, 8q, 12p, and 18p, were also identified and remained consistent as the disease advanced. EZH2 mutations were linked to a significant downregulation of histone genes, potentially contributing to tumor progression through epigenetic dysregulation [97]. Moreover, several clinical risk factors have been identified as key determinants of early disease progression. These risk factors include ECOG-PS, group B symptoms, age, male, LDH, β2-MG, hemoglobin, EZH2 mutation, Ann Arbor stage 3 or 4, FLIPI scores ranging from 2 to 5, high metabolic tumor volume, as well as increased lymphoma-associated macrophages [11, 17, 98–100]. Regarding MZL, in the study by Conconi et al. [49], patients with high-risk MALT-IPI were more likely to experience early POD (p = 0.006). The revised MALT-IPI outperformed the original in identifying high-risk POD24 patients, with an AUC of 0.734 compared to 0.684 for MALT-IPI [79]. MZL-IPI also demonstrated predictive value, with 6%, 15%, and 34% of early progressions occurring in low-, intermediate-, and high-risk groups, respectively [80]. Other factors associated with POD24 include age > 60, performance status, systemic symptoms, bone marrow involvement, and abnormal lab values (low albumin, elevated LDH, etc.) [48]. For CLL, early POD is linked to factors such as age ≥ 75, del(17p), non-FCR/BR first-line therapy, treatment duration ≤ 4 months and undetectable MRD (< 10^− 5) [60, 101].

Table 3.

Prognostic Models in Follicular Lymphoma and Correlation with POD24

Study	N		Accuracy	Sensitivity	Specificity	PPV	NPV
Jurinovic18	GLSG (N=132)/ BCCA (N=102)	FLIPI	67%/64%	78%/70%	56%/58%	27%/33%	92%/87%
		m7-FLIPI	70%/65%	61%/43%	79%/86%	38%/48%	91%/84%
		POD24-PI	73%/67%	78%/61%	67%/73%	33%/40%	94%/87%
Huet77	N = 488	23-gene predictor	-	43%	79%	38%	82%
Mir75	N = 1202	FLEX	-	60%	68%	-	-
		FLIPI	-	53%	58%	-	-
		FLIPI-2	-	53%	59%	-	-
		PRIMA-PI	-	69%	48%	-	-
Kuroki19	N = 45	High TLG and non-CR/CMR	89%	56%	100%	100%	87%
Hu89	N = 135	PRIMA-PI	54%	75%	50%
		PRIMA-PIC	56%	92%	48%
Bachy16	N = 1135		Low-risk	interm-risk	high-risk	P
		FLIPI	16%	21%	31%	1.36*10− 5	-
		PRIMA-PI	14%	21%	38%	1.41*10− 12	-
Tobin93			Low PD-L2	highPD-L2	P
	discovery cohort(N = 198)	PD-L2	45.70%	16.30%	0 .001
	GLSG2000(N = 138)		54.2%	14.30%	0.011
	BCCA(N = 45)		46.70%	24.00%	0.011

PPV positive predictive value, NPV negative predictive value, FLIPI follicular lymphoma international prognostic index, POD24-PI progression of disease within 24 months prognostic index, FLEX follicular Lymphoma Evaluation Index PRIMA-PI PRIMA-Prognostic Index TLG total lesion glycolysis, non-CR/CMR less than complete [metabolic] response

Moreover, Scoring systems like IPSS-WM, rIPSS-WM, MSS-WM, and CLL-IPI effectively distinguish PFS and may also help predict POD24, though further validation is needed. While advances in predictive modeling for POD24 have been made, no tool currently offers the ideal combination of sensitivity and specificity for reliably identifying high-risk patients at the onset of treatment. The search for more accurate predictive systems remains an area of active research, with the hope that further refinements in genomic and clinical risk assessment will eventually yield more precise tools to improve patient outcomes.

Discussion

The impact of POD24 has been consistently established in the context of most indolent lymphomas, substantiating its significance as a prognostic factor, particularly in follicular lymphoma. Despite this, research focusing on its implications in CLL/SLL remains limited, particularly in the era of novel agents. Further investigation is needed to confirm the consistency of findings and establish its prognostic significance in these specific subtypes. Similarly, the relationship between early POD and prognosis in patients with LPL/WM lacks substantial research evidence. Given the distinct characteristics and clinical behavior exhibited by LPL/WM compared to other indolent lymphomas, it is imperative to conduct dedicated studies to comprehensively delineate the prognostic implications associated with early disease progression within this patient cohort. Beyond indolent lymphomas, studies have confirmed that POD24 remains a strong predictor of prognosis in diffuse large B-cell lymphoma, peripheral T-cell lymphoma and mantle cell lymphoma [102–110]. Early POD is an emerging and important prognostic factor that may help to identify high-risk patients with an otherwise good-prognosis disease [111]. However, there is currently no risk stratification system that accurately identifies this group of patients. To better harness the complementary role of POD24 on survival, perhaps we can combine POD24 with initial traditional risk to formulate a novel prognostic classification incorporating both static and dynamic risk. This approach combines both static and dynamic risk factors, allowing for a more robust and holistic prognostic assessment. Therefore, we guess that combining the POD24 with baseline assessment tools can potentially enhance the guidance for clinical patient management in indolent lymphomas. In the meantime, using the most effective available treatment regimen, combining antibody therapy and chemotherapy, may help reduce the number of patients experiencing early relapse. Studies have shown that BR and Obinutuzumab-based regimens can reduce the incidence of POD24 and prolong PFS, while R-maintenance therapy can also extend PFS. However, there is currently no research on the relationship between R-maintenance therapy, POD24, and OS. It appears that for high-risk POD24 patients, these treatment options can be considered during initial therapy [13, 15, 35]. In addition, there is currently no standardized treatment protocol for patients experiencing early disease progression. Available treatment options include BTK inhibitors, PI3K inhibitors, hematopoietic stem cell transplantation, and CAR-T cell therapy [111–116].

The utilization of OS as a clinical trial endpoint for patients with indolent lymphoma characterized by prolonged survival periods is increasingly impractical. This is primarily due to the considerable length of time required to gather data on a statistically significant number of deceased patients. Considering the notable predictive capability of POD24 concerning poor prognosis in affected individuals, there is a growing inclination to propose its adoption as a novel trial endpoint [11, 23, 49]. Substituting OS with POD24 as an endpoint in clinical trials would offer several advantages, including a more efficient and timely evaluation of treatment outcomes, a greater likelihood of achieving meaningful results, and an improved ability to guide clinical decision-making for patients with indolent lymphomas.

Conclusion

Despite advancements in indolent lymphoma treatment, a subset of patients faces early relapse, often indicating a grim prognosis. While early disease progression lacks a standard definition, most studies adopt a 24-month timeframe, defining Progression of Disease at 24 months as progression, relapse, or death from lymphoma or treatment-related issues. Research consistently shows that POD24 correlates with lower survival rates and a notable subset experiences histological transformation.

Understanding the root causes of POD24 and histological transformation is crucial to tailor treatment and improve outcomes for this high-risk group. Better tools are needed to identify patients at risk of early progression. Future endeavors should prioritize research aimed at reducing POD24 incidence, optimizing prognosis, and determining the most effective treatment strategies for patients experiencing early disease progression to enhance overall care and outcomes in indolent B-cell non-Hodgkin lymphomas.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Xiaoyan Li wrote the main manuscript text and prepared all tables，Xin Wang provided ideas and took responsibility for revisions.

Funding

This work was supported by Natural Science Foundation of Chongqing Municipality, NO. cstc2021jcyj-msxmX0179.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.