Prognostic Impact of Serum β2-Microglobulin Levels in Hodgkin Lymphoma Treated with ABVD or Equivalent Regimens: A Comprehensive Analysis of 915 Patients

Simple Summary The significance of serum beta-2 microglobulin (sβ2m) in Hodgkin lymphoma (HL) is controversial. In an effort to investigate the prognostic significance of sβ2m levels in a large series of patients with HL, we analyzed 915 patients, who were treated with ABVD or equivalent regimens with or without radiotherapy. Sβ2m levels were measured by a radioimmunoassay (upper normal limit 2.4 mg/L). The median sβ2m levels were 2.20 mg/L. Freedom from progression (FFP) was significantly inferior in patients with a higher sβ2m at all tested cutoffs. The best cutoff was 2.0 mg/L (10-year FFP 83% vs. 70%, p = 0.001), which performed better than the 2.4 mg/L cutoff (“normal versus high”). Our data suggest that higher sβ2m is a significant independent predictor of FFP, OS and HLSS in HL but the optimal cutoff appears to lie within the normal limits in this predominantly young patient population. Abstract The significance of serum beta-2 microglobulin (sβ2m) in Hodgkin lymphoma (HL) is controversial. We analyzed 915 patients with HL, who were treated with ABVD or equivalent regimens with or without radiotherapy. Sβ2m levels were measured by a radioimmunoassay (upper normal limit 2.4 mg/L). Sequential cutoffs (1.8–3.0 by 0.1 mg/L increments, 3.5 and 4.0 mg/L) were tested along with ROC analysis. The median sβ2m levels were 2.20 mg/L and were elevated (>2.4 mg/L) in 383/915 patients (41.9%). Higher sβ2m was associated with inferior freedom from progression (FFP) at all tested cutoffs. The best cutoff was 2.0 mg/L (10-year FFP 83% vs. 70%, p = 0.001), which performed better than the 2.4 mg/L cutoff (“normal versus high”). In multivariate analysis, sβ2m > 2.0 mg/L was an independent adverse prognostic factor in the whole patient population. In multivariate overall survival analysis, sβ2m levels were predictive at 2.0 mg/L cutoff in the whole patient population and in advanced stages. Similarly, sβ2m > 2.0 mg/L independently predicted inferior HL-specific survival in the whole patient population. Our data suggest that higher sβ2m is an independent predictor of outcome in HL but the optimal cutoff lies within the normal limits (i.e., at 2.0 mg/L) in this predominantly young patient population, performing much better than a “normal versus high” cutoff set at 2.4 mg/L.


Introduction
The prognosis of Hodgkin lymphoma (HL) has dramatically changed over the last few decades, with the 5-year survival rate below 10% in the 1960s increasing to a 10-year survival rate exceeding 80% in the 2010s [1,2].A further increase is expected with the use of novel immunotherapies for relapsed/refractory disease [3][4][5][6] or even their incorporation in earlier treatment lines [7,8].The prognosis primarily depends on clinical stage as defined by the anatomic extent of the disease and the presence of B-symptoms according to the Ann Arbor staging system [9] and the Cotswolds [10] and Lugano modifications [11].In early-stage disease, the presence of bulky mediastinal disease, the number of involved nodal sites, elevated erythrocyte sedimentation rate (ESR), extranodal involvement and age provide additional prognostic information [12][13][14][15][16][17][18][19][20][21][22][23][24], while the 7-factor international prognostic score (IPS) has become the standard prognostic tool for advanced stages [25] followed by simplified versions [26,27].Another 7-factor advanced-stage Hodgkin lymphoma International Prognostic Index was recently published by the HoLISTIC consortium including significantly overlapping factors compared to IPS, albeit handled in a totally different logistic way [28].
Unfortunately, the above and other conventional prognostic systems cannot accurately classify patients with highly divergent levels of risk of relapse/progression and are unable to define either a very low-risk subgroup or any sizeable subgroup of patients with a>40-50% failure rate under modern ABVD-like fixed or Positron Emission Tomography driven therapy [25][26][27][28][29][30].Many biological prognostic factors have been evaluated in this context, but none has been adopted in everyday prognostication for several reasons [30].Thus, research is still focusing to the identification of novel, powerful, conventional and biological prognostic factors which might permit the reduction in chemotherapy and omission radiotherapy (RT) in an effort to minimize the long-term toxic effects in low-risk patients and guide the intensification/modification of treatment in high-risk groups.
Serum beta 2 -microglobulin (sβ 2 m) is a well-established prognostic factor in multiple myeloma, and has been incorporated in the international staging system (ISS) [31].It is also an extensively evaluated prognostic factor in diffuse large B-cell lymphoma and other non-Hodgkin lymphoma subtypes [32][33][34][35][36][37][38][39][40][41][42][43][44][45][46] and may work in acute myeloid leukemia as well [47], but has not been incorporated in current prognostic models for these diseases.Although tested as a prognostic factor in HL 30 years ago [48], its role has not yet been fully established; thus, sβ 2 m has not been used in any of the current prognostic systems for HL.However, there are several small-to medium-sized studies evaluating the potential prognostic role of sβ2m levels in patients with HL using heterogenous therapy, different endpoints, and various cutoffs with conflicting results [48][49][50][51][52][53][54][55][56].The data from our group, published in 2002 and 2005 based on patients treated with ABVD or equivalent regimens, were also conflicting [52,53].The updated analysis of 379 patients in 2005 demonstrated a role of sβ 2 m in predicting overall survival and also in predicting failure-free survival in the early stages only [53].
At this point, we extended our series to include 915 patients treated optimally with ABVD or equivalent regimens with or without RT with a much longer median follow-up of approximately 9 years extending up to almost 30 years.The size of this population is by far the largest ever recruited and permits the extraction of much more reliable conclusions.It also enables the reliable evaluation of multiple cutoffs, since the prognostic significance of sβ 2 m levels is not necessarily evident in a "normal versus elevated" analysis and optimal cutoffs for clinical use may also be different in different disease stages.

Patients, Staging, Treatment Strategies and Laboratory Assays
We analyzed 915 patients who received a diagnosis and first-line treatment for HL between 1990 and 2018, and had available sβ 2 m levels at diagnosis.The study period was extended from the beginning of sβ 2 m-level determination in clinical practice until the change in the method of sβ 2 m measurement implying a different cutoff in 2018.All patients were older than 14 years, were HIV-negative, and had received treatment with anthracycline-based CT with or without RT.In this retrospective study, patients were selected solely based on the availability of pretreatment sβ 2 m levels and their characteristics were comparable with those of patients who had also received anthracycline-based chemotherapy with or without RT during the same period, but did not have available serum β 2 -microglobulin levels, as previously reported [52,53].
All patients were clinically staged according to the Ann Arbor system [9], using standard staging procedures.Clinical Ann Arbor stages (AAS) IA and IIA were considered early, while clinical stages IB, IIB, III and IV were considered advanced for the purposes of this analysis.The number of involved anatomic sites was determined as described in previous publication of our group.Hemoglobin, white blood cell counts, and the differential erythrocyte sedimentation rate (ESR), serum albumin and serum LDH levels were measured by standard assays.Anemia was defined as the presence of hemoglobin levels <13 g/dL for males and <11.5 g/dL for females.Serum albumin was analyzed at a cutoff of 4 g/dL, as proposed by the IPS [25].Severe lymphopenia was also defined according to the cutoff provided by the IPS (<0.6 × 10 9 /L or <8%) [25].
Treatment strategies for early (IA, IIA) and advanced AAS (IB, IIB, III, IV) patients have been described previously [52,57].PET-driven strategies have been adopted during the last 15 years, initially for advanced-and later for early-stage disease.The evolution of treatment strategies during the study period has been described in recent publications of our group [58,59].
Sβ2m was measured using a radioimmunoassay (Pharmacia).The range of normal values was 1.0-2.4mg/L.
The study was approved by the appropriate Institutional Review Board.As a noninterventional retrospective study, informed consent was waived.

Statistical Analysis
The frequency of elevated sβ 2 m levels among various subgroups of patients were compared by the chi-square test.The Mann-Whitney and Kruskal-Wallis tests were used for non-parametric comparisons, as appropriate.The correlation between sβ 2 m levels and other variables evaluated as continuous was estimated by the Spearman's rho coefficient.The optimal cutoff for sβ 2 m levels was determined by direct testing of sequential cutoffs and by the use of Receiver Operator Curves (ROCs).The results obtained by both approaches were very similar.
Freedom from progression (FFP) was defined as the time interval between treatment initiation and treatment failure or last follow-up.Treatment failure was defined as the inability to achieve complete or partial remission (CR, PR) during initial therapy, requiring a switch to alternative chemotherapy, or relapse/progression after an initial CR/PR or toxic death.Patients with deaths of unrelated causes were censored.Overall survival (OS) and Hodgkin lymphoma-specific survival (HLSS) were measured from treatment initiation to death from any cause or HL-related causes (progressive HL, death of treatment toxicity), respectively, or last follow-up.Deaths due to secondary malignancies or cardiovascular causes during CR were censored.Survival after failure (SAF) was defined as the time interval between the documentation of treatment failure (primary failure or relapse) and death from any cause or last follow-up.The estimation of actuarial FFP or survival was performed using the Kaplan-Meier method [60].The identification of prognostic factors in univariate analysis was based on the log-rank test [61].The identification of independent prognostic factors was performed using Cox's proportional hazards model [62].
The correlation between sβ 2 m levels and other potential prognostic factors is shown in Table 1.Sβ 2 m levels correlated strongly with all baseline features, including demographics (older age, male gender), non-nodular sclerosing classical HL, clinical and laboratory markers of disease extent and aggressiveness and the IPS (all p-values < 0.001), with only correlations with leukocytosis, iliac/inguinal and lung involvement being looser but still statistically significant.With respect to potential biological prognostic factors, highly significant correlations of moderate magnitude were observed between sβ 2 m and serum soluble CD30, serum interleukin-10 and serum ferritin (p < 0.001 but Spearman's rho 0.333-0.455),as summarized in Table 2.However, there were no significant associations with bcl-2, activated caspase-3 or Epstein-Barr virus Latent Protein-1 (LMP-1) immunohistochemical expression.

Freedom from Progression
The 10-year FFP rate for the whole series was 76%.Among 208 events, only 3 were toxic deaths, while 205 were related to progressive or relapsing disease.As expected, most of the potential prognostic factors listed in Table 1 were statistically significant in the univariate analysis of FFP at the level of ≤0.001, with the exception of iliac/inguinal involvement (p = 0.047), age, gender, histology, leukocytosis, lung involvement and the number of nodal sites (in advanced disease only), which were not significant.

All Patients
When sβ 2 m levels were classified as quartiles, a consistent drop of 5-6% was observed for each one from Q1 to Q4, with 10-year rates of 84%, 78%, 73% and 68% (p = 0.001, Figure 1A), indicating a "dose-response" effect.Patients with elevated sβ2m levels had inferior FFP (70% versus 80%, p = 0.001, Figure 1B).As the "normal versus elevated" comparison is arbitrary and may not be optimal, several cutoff points for sβ2m levels were evaluated to identify the optimal cutoff to predict FFP, starting from 1.8 mg/L and advancing in 0.1 mg/L steps up to 3.0 mg/L and then at 0.5 mg/L steps up to 4.0 mg/L.In the univariate analysis, FFP was significantly inferior in patients with higher sβ2m at all tested cutoffs, as shown in Table 3.Interestingly, the Patients with elevated sβ 2 m levels had inferior FFP (70% versus 80%, p = 0.001, Figure 1B).As the "normal versus elevated" comparison is arbitrary and may not be optimal, several cutoff points for sβ 2 m levels were evaluated to identify the optimal cutoff to predict FFP, starting from 1.8 mg/L and advancing in 0.1 mg/L steps up to 3.0 mg/L and then at 0.5 mg/L steps up to 4.0 mg/L.In the univariate analysis, FFP was significantly inferior in patients with higher sβ2m at all tested cutoffs, as shown in Table 3.Interestingly, the 2.4 mg/L cutoff ("normal versus elevated") was not the best one, as the widest difference was observed at the cutoff of 2.0 mg/L(10-year FFP 83% versus 70%, p < 0.001; Figure 1C).ROC curve analysis confirmed this finding and provided a best cutoff at 2.02 mg/L.The area under the curve (AUC) was 0.573 (95% CI 0.53-0.62;p = 0.01).sβ 2 m levels >2.0 mg/L were an independent adverse prognostic factor in the largescale multivariate analysis (see Table 4 footnote) of all 915 patients, along with stage and lymphocytopenia (hazard ratio (HR) 1.55, 95% confidence intervals (CI) 1.11-2.17,p = 0.01; Table 4).The "normal versus elevated" comparison was not significant in the multivariate analysis (Table 4).

Early Stages
Among 515 patients with early-stage HL (IA/IIA), the best cutoff was found at 1.9 mg/L, with 10-year FFP rates of 88% versus 78% (p = 0.003, Figure 1D).Significant results were also obtained at the cutoff of 2.0 mg/L, with 10-year FFP rates of 86% versus 78% (p = 0.007, Figure 1E).As shown in Table 3, cutoffs set at 2.2 mg/L or higher, including the "normal versus elevated" comparison were not predictive of FFP.Sβ 2 m levels > 2.0 mg/L were an independent adverse prognostic factor in a large-scale multivariate analysis of patients with early stages along with ≥3 nodal sites and ESR ≥ 50 mm/h (hazard ratio (HR) 1.65, 95% confidence intervals (CI) 1.04-2.62,and p = 0.034; Table 4).Unexpectedly, the "normal versus elevated" comparison was also significant in multivariate analysis with a similar HR (Table 4).

Advanced Stages
Among 400 patients with advanced-stage HL (IB/IIB/III/IV), none of the tested cutoffs, including the "normal versus elevated" comparison, were predictive of FFP in univariate analysis (Table 3).The best cutoff was set at 2.0 mg/Land resulted in a marginally significant prediction, with 10-year FFP rates of 74% versus 64% (p = 0.09, Figure 1F).Similarly to the univariate results, sβ 2 m levels > 2.0 mg/L were an independent adverse prognostic factor of borderline significance in the multivariate analysis of patients with advanced stages, including all the IPS factors (Table 4; see also footnote) along with stage IV, lymphopenia and leukocytosis (protective!!) (hazard ratio (HR) 1.44, 95% CI 0.94-2.21,and p = 0.098; Table 4).The "normal versus elevated" comparison was not significant in multivariate analysis (Table 4).

Overall Survival
The 10-year OS rate for the whole series was 85%.Among 131 deaths, 74 were diseaserelated and 57 unrelated.

Early and Advanced Stages
In the multivariate analysis of OS in early-stage patients, sβ 2 m levels were neither an independent prognostic factor at the cutoff of 2.0 mg/L nor at a "normal versus elevated" basis (Table 4).In contrast, in advanced stages, sβ 2 m > 2.0 mg/L was an independent adverse prognostic factor along with older age, lymphopenia, anemia and a lack of leukocytosis (hazard ratio (HR) 2.07, 95% CI1.04-4.15,and p = 0.039; Table 4).The "normal versus elevated" comparison was not predictive in advanced-stage disease (Table 4).

Causes of Death, Hodgkin Lymphoma-Specific Survival and Survival after Failure
Up to publication of this study, 131 deaths were recorded.Among them, 74 (56%) were due to HL, with 58 being directly related to progressive HL, 5 toxic deaths (3 during first-line and 2 during salvage therapy), 9 secondary neoplasias plus active HL and 1 congestive heart failure directly after treatment.We also recorded 27 unrelated deaths and 30 deaths of secondary neoplasia during first complete remission.
In the multivariate analysis of HLSS in early-stage patients, sβ 2 m levels were the only variable with a borderline-independent effect, only when evaluated at the cutoff of 2.0 mg/L (hazard ratio (HR) 2.30, 95% CI 0.89-5.94,and p = 0.085; Table 4), but not in a "normal versus elevated" basis (Table 4).In contrast, in the multivariate analysis of HLSS in advanced stages, sβ 2 m levels were neither an independent prognostic factor at the cutoff of 2.0 mg/L nor at a "normal versus elevated" basis (Table 4).

Discussion
B 2 m is synthesized in all nucleated cells, binds to major histocompatibility complex class I molecules, and is not directly attached to the cell membrane.Thus, free soluble β 2 m is detected in body fluids due to its release from the cell surface and cytoplasm.Since its identification 50 years ago, in 1972 [63], β 2 m has been widely investigated as a prognostic factor in hematologic malignancies.Further to the correlation with tumor burden [52], the mechanisms underlying the prognostic significance of β 2 mstill remain unclear.Indeed, several studies have shown that the prognostic significance of sβ 2 mmay be independent from factors reflecting disease burden [33,35], suggesting that it could either be related to other specific biologic features of lymphomas or simply overcome other markers of tumor burden, obscuring their significance.
The role of sβ 2 m as a prognostic factor in HL has been evaluated in several smallor medium-sized studies in the past, with partially conflicting results, probably owing to the sample sizes and the variable treatment approaches (Table 5) ( [48][49][50][51]53,55,56,[64][65][66][67][68][69], present study).Briefly, considering both the MD Anderson studies together [48,51], sβ 2 m appears to correlate with FFS in advanced stages (overall survival not reported), while it was associated with inferior overall survival in early stages, with only a borderline effect on FFS.It should be noted that treatment was inferior to ABVD and equivalents in the majority of these patients.In a study of the International Hodgkin Study Group, sβ 2 m was an independent predictor of FFS in early stage patients with favorable characteristics treated with RT alone [66].In the present study, we evaluated the prognostic role of sβ 2 m in HL in the-by far-largest series published to date, consisting of a large cohort of homogeneously treated patients.Importantly, all patients had been treated with ABVD or equivalent regimens with or without RT, which are considered standard therapy for HL.Since it is known that more effective treatment may eliminate the significance of previously established prognostic factors, our study rules out a potential bias due to inferior treatment.Our results extend our previous observations and establish sβ 2 m as a potential independent prognostic factor in HL.
An important novel observation was made possible thanks to the very large size of this patient population: analyzing several potential cutoffs, we concluded that sβ 2 m may not work well as a prognostic factor, neither when analyzed on a "normal versus elevated" basis at the cutoff of 2.4 mg/L(as performed in our previous studies) [52,53] nor when analyzed at higher-clearly abnormal-cutoffs, as 2.5 mg/L [48,51,55,56] or 3.0 mg/L [65].Instead, sβ 2 m worked better when the cutoff was set within the normal range at 2.0 mg/L.It is reasonable to wonder whether this observation is biologically relevant.In our opinion it is reasonable, because sβ 2 m levels are strongly and positively correlated with age in normal subjects.As the normal range is established from unselected normal individuals from the general population, the true upper normal limit for younger people might probably be lower.Along these lines, as patients with HL are much younger than the general population, the expected upper normal limit of sβ 2 m for the majority of them might be lower than the conventional 2.4 mg/L and might approach 2.0 mg/L.
The use of a sβ 2 m cutoff within the normal range of is supported by a recent Chinese study, in which 353 patients were evaluated, among whom 230 had received ABVD and 123 ABVD-like regimens, the latter with inferior progression-free survival (PFS).The levels of sβ 2 m were evaluated by ROC curves and the best cutoff was set at 1.85 mg/L, very similarly to our results.Although sβ 2 m levels above that cutoff were associated with inferior PFS and OS in the univariate analysis, the prognostic significance was independent of other factors only for OS.This is not unexpected, as sβ 2 m levels are more potent predictors of OS, as shown in the present and our previous studies, because of their strong association with age and renal function.However, the moderate size of the study by Wen et al. might have obscured an independent effect of sβ 2 m levels on PFS.In another small study of 67 patients, ROC analysis suggested a cutoff of 2.5 mg/L, which produced significant results in multivariate analysis for PFS, OS and DSS [55].
Reporting here our experience in 915 patients with HL, with 208 treatment failure events recorded so far (84 in early and 124 in advanced stages), this study was powered to detect moderate but clinically significant differences and to perform subgroup analyses according to clinical stage.Serum β 2 mlevels >2.0 mg/L independently predicted a lower FFP rate in the whole-patient population of this study, when evaluated in multivariate analysis including 11 additional and potentially strong prognostic covariates.The same was true for OS and HLSS.Notably, on a "normal versus elevated" basis, sβ 2 m had no independent effect on FFP and HLSS, presenting only a borderline association with OS.Among 515 patients with early-stage HL (IA/IIA), the best cutoff was found at 1.9 mg/L, but 2.0 mg/L was also highly significant and was used for further evaluation for reasons of consistency.Again, sβ 2 m was an independent predictor of FFP, when evaluated in multivariate analysis including eight additional covariates with established or strongly suspected prognostic significance in early-stage disease.Serum β 2 m levels were the only independent predictor of HLSS, albeit of marginal significance, but had no effect on OS.Among 400 patients with advanced-stage HL (IB/IIB/III/IV), sβ 2 m levels >2.0 mg/L were an independent predictor of FFP, when evaluated in multivariate analysis including all seven IPS covariates.This effect was more marked regarding OS but sβ 2 m had no independent effect on HLSS.On a "normal versus elevated" basis, sβ 2 m had no independent effect in any of the three endpoints.
Finally, this is the first study to show that sβ 2 m levels are significantly correlated with some established biological prognostic factors, such as serum sCD30 and serum IL-10 levels [66], but not with others, such as the tissue expression of bcl-2 and activated caspase-3 [30].

Conclusions
The present study has established the prognostic impact of sβ 2 m at a lower-thanexpected cutoff, but also raises several new questions.It is not clear if sβ 2 mcan add to the prediction achieved by the IPS or other prognostics systems or just replace variables within the existing systems.Unfortunately, sβ 2 m was not evaluated during the development of the new holistic IPS [28], while its additive impact when biological prognostic factors are taken into account remains unknown.Some kind of "correction" according to the renal function should also be investigated in the effort to increase the prognostic significance of sβ 2 m levels.Another problem is that interim PET-guided therapy has predominated the field of treatment of Hodgkin lymphoma during the last decade.It is not clear whether sβ 2 m simply predicts a higher probability of interim PET positivity or may further help to discriminate which patients with a negative interim PET will relapse or who will be cured with intensified treatment following a positive interim PET.Similar considerations apply regarding the potential association of sβ 2 m levels with the results of end-of-treatment PET [70].In addition, as novel prognostic factors appear, their correlation with sβ 2 m should be accurately determined.Unfortunately, there are no data regarding the correlation of sβ 2 m, neither with the circulating tumor DNA [71][72][73] and its changes during treatment nor with PET metrics, including baseline total metabolic tumor volume (TMTV) [74][75][76][77][78], total lesion glucolysis (TLG) [79,80] or lesion dissemination [81][82][83].As sβ 2 m levels correlated strongly with almost all baseline features reflecting disease extent and aggressiveness in this study, it is reasonable to hypothesize a strong correlation with PET metrics, which, however, does not exclude the persistence of the independent prognostic significance of sβ 2 m.Finally, probably the main question to be asked in the near future is how sβ 2 m will affect the outcome of patients treated in the first line with chemotherapy plus novel agents such as BV-AVD or BreCADD, or-more importantly-how sβ 2 m will work as a prognostic factor under treatment with nivolumab-AVD [8].

Figure 2 .Figure 2 .
Figure 2. (A) Overall survival (OS) according to sβ2m levels classified as quartiles in the whole patient population; (B) OS according to sβ2m levels as "normal vs. elevated" in the whole patient population (≤2.4 mg/L vs. >2.4mg/L);(C) OS according to sβ2m levels (≤2.0 mg/L vs. >2.0mg/L) in the whole patient population; (D) Hodgkin lymphoma-specific survival (HLSS) according to sβ2m levels classified as quartiles in the whole patient population; (E) HLSS according to sβ2m levels as "normal vs. elevated" in the whole patient population (≤2.4 mg/L vs. >2.4mg/L); (F) HLSS according to sβ2m levels (≤2.0 mg/L vs. >2.0mg/L) in the whole patient population; (G) survival after failure (SAF) Figure 2. (A) Overall survival (OS) according to sβ 2 m levels classified as quartiles in the whole patient population; (B) OS according to sβ 2 m levels as "normal vs. elevated" in the whole patient population (≤2.4 mg/L vs. >2.4mg/L); (C) OS according to sβ 2 m levels (≤2.0 mg/L vs. >2.0mg/L) in the whole patient population; (D) Hodgkin lymphoma-specific survival (HLSS) according to sβ 2 m levels classified as quartiles in the whole patient population; (E) HLSS according to sβ 2 m levels as "normal vs. elevated" in the whole patient population (≤2.4 mg/L vs. >2.4mg/L); (F) HLSS according to sβ 2 m levels (≤2.0 mg/L vs. >2.0mg/L) in the whole patient population; (G) survival after failure (SAF) according to sβ 2 m levels classified as quartiles in the whole patient population; (H) SAF according to sβ 2 m levels as "normal vs. elevated" in the whole patient population (≤2.4 mg/L vs. >2.4mg/L); and (I) SAF according to sβ 2 m levels (≤2.0 mg/L vs. 2.0 mg/L) in the whole patient population.

Table 3 .
Univariate analysis of the prognostic significance of serum β 2 -microglobulin levels on various cutoff points in terms of freedom from progression.

Table 4 .
Multivariate analysis of the prognostic significance of serum β 2 -microbulin levels of freedom from progression, and overall and Hodgkin lymphoma-specific survival.Analysis performed at the cutoff of 2.0 mg/L or a "normal vs. elevated" basis (cutoff 2.4 mg/L) in the whole potent population and in early and advanced stages separately.

Table 5 .
Summary of published studies on the prognostic significance of serum β 2 -microglobulin levels in patients with Hodgkin lymphoma.Not Applicable, NT = Not Tested, NR = Not Reported, ES = early stage, non-X = non bulky.* Tumor control was the endpoint either as TTF or FFP (with progression/relapse counted as events along with treatment-or disease-related deaths or not; deaths of any other cause were censored).** RFS was the endpoint (only recurrence counted as event).*** DFS and cause-specific survival were the endpoints.§ 111/167 had sβ 2 m levels available; ABVd = ABVD with reduced dacarbazine doses.§ § Cutoff determined by ROC curve analysis.¶ Independent prognostic factor for achievement of CR but not for disease-free survival.¶ ¶ The effect of sβ 2 m within early and advanced stages was tested only in univariate analysis.