Prognostic Utility of a Novel Prognostic Model Consisting of Age, CRP, Ki67, and POD24 in Patients with Angioimmunoblastic T-Cell Lymphoma

Abstract

To find the independent factors affecting the prognosis of AITL patients, establish a novel predictive model, and stratify the prognosis of AITL patients. We retrospectively analyzed the clinical data of 86 patients diagnosed with AITL in the First Affiliated Hospital of Wenzhou Medical University from December 2010 to March 2022. The clinical features, recurrence time, and death time of patients were collected and analyzed statistically. The median age of our patients was 68 years old, and the male-to-female ratio was 2.2: 1. There are differences between males and females in ECOG PS score (p = 0.037), β2 microglobulin levels (p = 0.018) and IgM (p = 0.021). Multivariate COX regression analysis showed that C-reactive protein > 39.3 mg/L (hazard ratio (HR), 5.41; p = 0.0001), Age > 66 years (hazard ratio (HR), 3.06; p = 0.0160), Ki67 positive (hazard ratio (HR), 4.86; p = 0.0010) and early progression of disease within 24 months (POD24) after diagnosis (hazard ratio (HR), 12.47; p = 0.0001) were independent factors affecting the prognosis of OS. KM analysis showed that the predictive model established by these four factors could effectively predict the prognosis of patients with AITL (p < 0.0001), and the ROC curve showed that the predictive ability of the new predictive model (AUC = 0.909) was significantly better than that of the traditional predictive models, such as IPI (AUC = 0.730), PIT (AUC = 0.720), PIAI (AUC = 0.715) and AITL score (AUC = 0.724). Age, C-reactive protein, Ki67, and POD24 were independent factors affecting the prognosis of OS. The prognostic model established by them combined clinical features, and serological and pathological indicators and could effectively predict the prognosis of AITL patients.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12288-024-01767-1.

Introduction

Angioimmunoblastic T-cell lymphoma is a relatively common T-cell lymphoma, accounting for approximately 1–2% of non-Hodgkin's lymphoma (NHL) and 15–20% of PTCL [1]. The median age of patients with AITL is 65, which is more common in Europe than in Asia [2]. AITL has an invasive course, and the specific clinical manifestations are hepatosplenomegaly, hypergammaglobulinemia, rash, and anemia [3, 4]. Pathologically, AITL is characterized by monoclonal T-cell infiltration and neovascularization [5]. AITL originated from T-follicular helper (TFHs) cells, which is associated with its unique clinical and pathological features [6, 7], CD10, CXCL13, BCL6, PD-1/CD279, CXCR5, ICOS, and CD154 are mostly positive [8–10]. Although the cause of AITL is unclear, there is growing evidence that EB virus infection is associated with AITL. 80–95% of AITL patients are positive for EBV in biopsies [11].

The prognosis of AITL is inferior. The 5-year overall survival (OS) rate is only 32%. In a retrospective study from 1990 to 2002, there was no difference in prognosis among patients treated with or without anthracycline [12]. However, a prospective cohort study from 2010 to 2014 showed that the use of anthracyclines, such as cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) regimens, could effectively improve prognosis [13]. Some studies have also suggested that adding etoposide to the CHOP regimen can improve the PFS of patients [14, 15]. Autologous hematopoietic stem cell transplantation (ASCT) after the first complete remission (CR1) can improve prognosis, according to a prospective study published in 2021 [16].

The International Prognostic Index (IPI) is a commonly used prognostic model, but the ability to identify high-risk PTCL patients is sometimes poor [17]. PTCL-U prognostic index (PIT) [18], the Prognostic Index for AITL(PIAI) [3], and the AITL score [16] were proposed, hoping to make up for the deficiency of IPI. However, there is still no clear prediction standard [19]; developing a better prognostic model would be helpful. This study aims to analyze the related factors affecting the prognosis of AITL and construct a new predictive model.

Materials and Methods

Patients

This study collected the clinical data of 86 AITL patients treated in the first affiliated Hospital of Wenzhou Medical University from December 2010 to March 2022. All patients were pathologically diagnosed as AITL, and the inclusion criteria were as follows: (A) pathologically proved AITL according to the WHO Classification of Tumors of Haematopoietic and Lymphoid Tissues, (B) no anti-tumor therapy before diagnosis, (C) with follow-up > 1 month. Ann Arbor staging criteria were used for staging. We follow up with the patients by consulting the hospitalization medical records and making phone calls. Overall survival (OS) was calculated as the interval between diagnosis and death of any cause or the last follow-up. The progression-free survival rate (PFS) was calculated as the interval between diagnosis and first recurrence, progression, or death.

Parameters

To establish a prognostic prediction model. We included the following parameters: Variables in IPI, PIT, and PIA (Age, Ann Arbor stage, performance status (ECOG), extranodal invasion, lactate dehydrogenase (LDH)、thrombocytopenia, B symptoms, and the presence or absence of bone marrow infiltration), variables related to immune status (serum IgA, IgG, and IgM levels, monocyte and lymphocyte count), previously proposed prognostic factors (white blood cell count, hemoglobin, serum albumin, C-reactive protein (CRP), Serum β-2 microglobulin levels), Lymph node biopsy (Ki67 positive index, EB virus-encoded small RNA (EBER) expression detected by in situ hybridization, Whether there is early disease progression within 24 months(POD24) after diagnosis [16]). The prognostic nutritional index (PNI) [20] was obtained by calculation, and the calculation formula is PNI = 10 × albumin (g/dL) + 0.005 × lymphocyte count (/mm3). These data were the patients' information at the time of initial diagnosis.

Statistical Analysis

Continuous variables were converted into classified variables through X-Tile programs (Yale University, New Haven, CT, USA) [21] and time-dependent receiver operating characteristic (ROC) curve analysis. Comparisons of clinical and prognostic features in males and females were performed using the chi-squared test or Fisher’s exact test. The survival differences among the stratified prognostic factors were analyzed by the Kaplan–Meier method and Log-rank test. The effects of risk factors on survival were analyzed by univariate and multivariate Cox proportional hazard regression analysis(P < 0.05). Multivariate COX analysis was used to identify variables that predict prognosis. The best prediction variable set was obtained by forward stepwise regression, and the optimal model was evaluated by Akaike Information Criteria (AIC). The results were presented as risk ratio (HR) and confidence interval (CI) of 95%. The nomogram represents the statistical prediction model through a simple graph [22]. It was constructed by the final COX regression model and was used to predict the 2-and 3-year OS rates of patients. The scores of each parameter of the final prediction model were added and converted to the probability of patients surviving at 2 and 3 years. Statistical analyses were performed with R (http://www.R-project.org), IBM SPSS Statistics (V19.0, IBM Corp., Armonk, NY), and EmpowerStats software (www.empowerstats.com, X&Y Solutions, Inc.Boston MA).

Results

Clinical Characteristics of AITL Patients

The clinical characteristics of 86 AITL patients included in the criteria were summarized in Table 1. The median age was 67.5 years (range 31–89 years), and 44 cases (51.2%) were over 66 years old. The male/female ratio was 2.2:1. 55 cases (64.0%) had B symptoms, 74 cases (86.0%) with Ann Arbor stage III ~ IV, HGB < 114 g/L in 44 cases (51.2%), 74 patients with low albumin level (86.0%), LDH > 299 U/L in 62 cases (72.1%), 32 cases (37.2%) with elevated IgG level, 16 cases (18.6%) with elevated IgA level, 24 cases (27.9%) with elevated IgM level, 49 cases (57.0%) with KI67 ≥ 50%, and 63 cases (73.3%) with EBER positive. IPI score ≥ 2 in 77 cases (89.5%), PIT score ≥ 2 in 60 cases (69.8%), and PIAI score ≥ 2 in 61 cases (70.9%). The patients those unable to evaluate POD24 due to loss of follow-up or death without disease progression within 24 months were excluded. POD24 occurred in 21 patients (33.9%), while 41 patients were without disease progression within the first 24 months (66.1%). Among all the clinical features, there were differences in IgM (p = 0.021), B2M levels (p = 0.018), and ECOG (p = 0.037) between males and females.

Table 1.

Patients Characteristics

Parameter		Female N.(%)	Male N.(%)	P-value
Age	> 66 years	15 (55.56%)	29 (49.15%)	0.581
ECOG PS score*	≥ 2	13 (48.15%)	15 (25.42%)	0.037
LDH*	> 299 U/L	21 (77.78%)	41 (69.49%)	0.427
B symptoms*	+	20 (74.07%)	35 (60.34%)	0.218
Ann Arbor Stage	III–IV	23 (85.19%)	51 (86.44%)	0.876
WBC*	> 7440/μL	15 (55.56%)	28 (47.46%)	0.486
Neutrophil	> 5225/μL	26 (55.32%)	17 (43.59%)	0.816
Monocyte	> 670/μL	13 (48.15%)	28 (47.46%)	0.953
Lymphocyte	> 1005/μL	17 (62.96%)	26 (44.07%)	0.104
Hemoglobin	< 114 g/L	16 (59.26%)	28 (47.46%)	0.310
Platelet	< 125 × 109/L	4 (14.81%)	15 (25.42%)	0.271
Albumin	< 40 g/L	24 (88.89%)	50 (84.75%)	0.607
IgG*	> 15.6 g/L	10 (37.04%)	22 (37.29%)	0.982
IgA*	> 4.53 g/L	3 (11.11%)	13 (22.03%)	0.227
IgM*	> 3.04 g/L	12 (44.44%)	12 (20.34%)	0.021
CRP*	> 39.3 mg/L	8 (42.11%)	18 (38.30%)	0.774
B2M levels*	> 2.5 mg/L	18 (66.67%)	52 (88.14%)	0.018
EBER	+	18 (66.67%)	45 (76.27%)	0.350
Ki67	≥ 50%	18 (66.67%)	31 (52.54%)	0.220
No. of extranodal sites	≥ 2	2 (7.41%)	3 (5.08%)	0.669
POD24*	+	6 (33.33%)	15 (34.09%)	0.954
PNI*	> 33.6	15 (55.56%)	24 (40.68%)	0.198
IPI*	≥ 2	25 (92.59%)	52 (88.14%)	0.531
PIT*	≥ 2	21 (77.78%)	39 (66.10%)	0.274
PIAI*	≥ 2	19 (70.00%)	42 (71.19%)	0.452

^*ECOG PS score: Eastern Cooperative Oncology Group performance status score; LDH: lactate dehydrogenase; WBC: white blood cell; IgG: immunoglobulin G; IgA: immunoglobulin A; IgM: immunoglobulin M; B2M: β2 microglobulin; CRP: C-reactive protein; POD24:early progression of disease within 24 months after diagnosis; PNI:prognostic nutritional index; IPI:the International Prognostic Index; PIT: Prognostic Index for Peripheral T-Cell Lymphoma, Unspecified; PIAI: the Prognostic Index for AITL

Cut-Off Points Identification of Clinical Continuous Variables

To convert continuous variables into taxonomic variables, to facilitate subsequent statistics, we used the following methods to define the cutoff values of clinical data in Table 1.

The area under the ROC curve (AUROC) of age is 0.68, and the best cut-off value is 66 years old (sensitivity, 67.6%; specificity, 64.6%). The AUROC value of LDH is 0.64, and the best cut-off value is 299U/L (sensitivity, 87.9%, specificity, 43.5% Personality 0.079). The AUROC value of CRP is 0.57, and the best cut-off value is 39.3 mg/L (sensitivity, 46.0%, specificity, 68.9% orientation 0.037, Supplementary Fig. 1).

According to the X-Tile program, when 114 g/L is used as the best cut-off point for hemoglobin, the maximum number of chi-square root points is 10.459 respectively. According to the distribution of hemoglobin at the time of diagnosis, the patients were divided into two groups: < 114 g/L and ≥ 114 g/L (P < 0.05, Supplementary Fig. 2A). When 33.6 was used as the best cut-off point for PNI, the maximum number of chi-square root points was 9.039. The patients were divided into two groups: ≤ 33.6 and > 33.6 (P = 0.0579, Supplementary Fig. 2B).

The cut-off values of IgA, IgM, IgG, and β2microglobulin levels are the upper limit of the normal range. The cut-off value of hemoglobin and platelet is the lower limit of the normal range. We used the median of included data for the cut-off values of white blood cell count, monocyte count, neutrophil count, lymphocyte count, and Ki67 positive index.

Treatment Outcomes and Survival

All patients received multidrug chemotherapy as the first-line regimen, and the initial chemotherapy regimens were CHOP or CHOP-like (n = 73), Gemox (n = 6), and MINE (n = 7). After treatment, 52 patients (73.2%) got CR/PR, and 19 (26.8%) got SD/PD. The end event ~ death ~ was observed in 45 patients(52.3%). The median OS and the median PFS were 23 and 17 months. 3-and 5-year OS rates were 41% (95% CI, 30% ~ 56%) and 36% (95% CI, 25% ~ 51%). 3and 5-year PFS rates were 37% (95% CI, 27% ~ 52%) and 30% (95% CI, 19% ~ 45%). Both IPI and PIT can predict OS (p < 0.001, Supplementary Fig. 3A-B).

Prognostic Factors

Univariate analysis showed that Age, ECOG, LDH, Ann Arbor stage, Hemoglobin, CRP, EBER, Ki67, POD24, and PNI had significant effects on OS (P < 0.05, Table 2), patients with thrombocytopenia tended to have shorter OS (p = 0.0573, Table 2). Age, ECOG, LDH, B symptoms, Ann Arbor stage, Hemoglobin, Platelet, CRP, β2 microglobulin, and EBER significantly affected PFS (P < 0.05, Supplementary Table 1). Multivariate analysis showed that POD24 positive(HR, 12.47; 95%CI, 4.00–38.93; p = 0.0001), CRP > 39.3 mg/L(HR, 5.41; 95%CI, 2.24–13.05; P = 0.0001), Ki67≧50%(HR, 4.86; 95%CI, 1.95–12.10; P = 0.0010), Age > 66 years old (HR, 3.21; 95%CI, 1.23–7.59; P = 0.0160) were independent poor prognostic factors for OS (Table 2). Age > 66 years old (HR, 3.06; 95%CI, 1.23–4.80; P = 0.0104)、CRP > 39.3 mg/L(HR, 2.65; 95%CI, 1.38–5.09; P = 0.0034)、Platelet < 125 × 109/L (HR, 2.81; 95%CI, 1.29–6.14; P = 0.0095) were independent poor prognostic factors for PFS (Supplementary Table 1). The prognostic effects of these significant factors were shown in the forest plots as follows (Fig. 1A-B).

Table 2.

Univariate and Multivariate Analysis of Prognostic Factors for OS

Parameter		Univariate analysis			Multivariate analysis
		HR	95%CI	p Value	HR	95%CI	p Value
AGE	> 66 years	2.71	1.45–5.08	0.0018	3.06	1.23–7.59	0.0160
Sex	Male	1.03	0.54–1.97	0.9205
ECOG PS score*	≥ 2	2.47	1.32–4.61	0.0047
LDH*	> 299 U/L	3.08	1.43- 6.67	0.0042
B symptoms*	+	1.77	0.94–3.34	0.0772
Ann Arbor Stage	III–IV	2.87	1.02–8.04	0.0454
Hemoglobin	< 114 g/L	2.30	1.27–4.19	0.0063
Platelet	< 125 × 109/L	1.93	0.98–3.82	0.0573
CRP*	> 39.3 mg/L	2.10	1.12–3.95	0.0207	5.41	2.24–13.05	0.0001
WBC*	> 7440/μL	1.52	0.83–2.75	0.1720
Neutrophil	> 5225/μL	1.29	0.72–2.33	0.3886
Monocyte	> 670/μL	0.65	0.36–1.17	0.1504
IgG*	> 15.6 g/L	1.03	0.56–1.91	0.9229
IgA*	> 4.53 g/L	0.87	0.39–1.96	0.7381
IgM*	> 3.04 g/L	0.88	0.45–1.74	0.7189
B2M levels*	> 2.5 mg/L	2.28	0.90–5.80	0.0828
EBER	+	2.35	1.09–5.08	0.0292
KI.67	≥ 50%	2.17	1.15–4.09	0.0170	4.86	1.95–12.10	0.0010
No. of extranodal sites	≥ 2	1.32	0.40–4.29	0.6484
PNI*	> 33.6	0.42	0.23–0.78	0.0062
POD24	+	3.01	1.53–5.92	0.0010	12.47	4.00–38.93	0.0001

^*ECOG PS score: Eastern Cooperative Oncology Group performance status score; LDH: lactate dehydrogenase; WBC: white blood cell; IgG: immunoglobulin G; IgA: immunoglobulin A; IgM: immunoglobulin M; B2M: β2 microglobulin; CRP: C-reactive protein; PNI:prognostic nutritional index; POD24:early progression of disease within 24 months after diagnosis

Fig. 1.

Forest plots of multivariate analysis for (A) OS and (B) PFS

Prognostic Model

Based on the multiple COX regression analyses, we tried to construct a new prognostic model, including age, CRP, Ki67, and POD24. We assigned each of the above four parameters 1 point and added them together to get the final total score. Patients were divided into three groups based on the total score: 0–1 for low-risk patients, 2 for Intermediate-risk patients, and 3–4 for high-risk patients. Kaplan–Meier curves showed that the new prognostic model could stratify the prognosis of OS in patients with AITL (P < 0.0001, Fig. 2). In addition, we used the final COX model for OS to construct a nomogram to predict patient survival (Fig. 3A). By adding up the scores of the patient's age, CRP Ki67, and POD24, the corresponding 2-year, and 3-year OS rates can be found directly in the chart.

Fig. 2.

Survival of patients with aitl according to the new prognostic model. this prognostic model could efficiently stratify the outcomes into 3 groups: patients with 0–1 risk factor (low risk), patients with 2 risk factors (intermediate-risk), and patients with 3–4 risk factors (high-risk)

Fig. 3.

(A)A Nomogram to predict the 2- and 3-year OS rates in AITL patients by Age, CRP, Ki67, and POD24.(B) AUC of the different models in predicting 2-year OS

Prognostic Power of the Novel prognostic Model

To evaluate the prognostic ability of this new predictive model in patients with AITL, we compared it with four existing AITL prognostic models by ROC curve. Figure 3B shows that our new prognostic model (AUC = 0.909) has a larger area under the curve (AUC) than the traditional models: IPI (AUC = 0.730), PIT (AUC = 0.720), PIAI (AUC = 0.715) and AITL score (AUC = 0.724), indicating better predictive value.

Discussion

AITL is a subtype of PTCL, which has a poor prognosis and is most common in the elderly. This study's median age of onset was 68 years old, similar to that previously reported [4]. Most AITL patients showed B symptoms, elevated LDH levels, advanced disease, and elevated immunoglobulin levels [23]. Induction chemotherapy for AITL patients is mainly based on CHOP or CHOP-like regimens. However, the efficacy of this anthracycline-based regimen is still controversial [12, 13]. Adding etoposide to CHOP can improve the response rate of induction chemotherapy [15]. The increase in the response rate of induction chemotherapy enables more patients to accept ASCT and improves the prognosis of patients. In our study, the patient's 5-year OS and 5-year PFS rates were 36% and 30%. It is similar to the results of recent studies (5-year OS and 5-year PFS rates were 44% and 32%, respectively).

The prognostic factors of AITL are still controversial. The International Prognostic Index (IPI), mainly used to predict the prognosis of B-cell lymphoma, is also a commonly used predictive model in AITL. However, it is controversial to identify high-risk PTCL patients [17]. In recent years, other predictive models have been proposed, such as PTCL-U prognostic index [18] and the Prognostic Index for AITL(PIAI) [3]. In the study of prognostic indicators of AITL, age, ECOG, Ann Arbor stage, multiple extranodal sites involved, leukocytosis, anemia, elevated LDH level, elevated β2 microglobulin level, elevated CRP level, elevated IgA level and serous cavity effusion may affect the survival of patients [16, 24, 25]. We included the factors that may be related to the prognosis of AITL in the study, and the results of univariate COX regression analysis were similar to those previously reported. The prognostic nutritional index (PNI), calculated from serum albumin and the number of peripheral blood lymphocytes, indicates nutritional status and inflammatory response. PNI has been proven to be a prognostic indicator of diffuse large B cell lymphoma(DLBCL) 、follicular lymphoma(FL), and extranodal natural killer/T cell lymphoma (ENKTL) [26–29]. In the study of our patients, in the univariate regression analysis affecting OS, the difference between the high and low PNI groups was statistically significant. However, after controlling multiple confounding factors, PNI could not independently affect the prognosis of AITL. Our current sample size is small, and subsequent follow-up and inclusion of new patients to increase the sample size may lead to more significant differences and the discovery of other parameters affecting the prognosis of AITL. C-reactive protein (CRP) is associated with interleukin-6 (IL-6) and plays an important role in lymphoma. Many studies have found that serum CRP levels affect the prognosis of aggressive NHL and Hodgkin lymphoma [30–32]. Ki-67 is a nuclear antigen in proliferative cells, representing the proliferative activity of tumor cells, and is associated with poor prognosis of lymphoma [3, 33, 34]. POD24 means patients with early progression of disease within 24 months after diagnosis. In 2021, Ana et al. found that POD24 was a strong prognostic factor in patients with AITL [16]. However, they did not include POD24 in their prognostic models. We tried it in this study and found that the effectiveness of the prediction model was significantly improved after the inclusion of the POD24 variable. In our study, Age, CRP, Ki67, and POD24 are independent prognostic factors in patients with AITL. The predictive model based on these four parameters has excellent predictive value. This new prediction model combines clinical, serological, and pathological characteristics of patients and is expected to guide clinical practice.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Competing Interests

The authors have declared that no competing interest exists.