The clinical value of tumor burden at diagnosis in Hodgkin lymphoma

Authors

  • Paolo G. Gobbi M.D.,

    Corresponding author
    1. Medicina Interna e Oncologia Medica, Universitá di Pavia, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico S. Matteo, Pavia, Italy
    • Medicina Interna e Oncologia Medica, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico S. Matteo, P. le Golgi no. 2, 27100 Pavia, Italy
    Search for more papers by this author
    • Fax: 011 (39) 0382526223

  • Chiara Broglia M.D.,

    1. Medicina Interna e Oncologia Medica, Universitá di Pavia, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico S. Matteo, Pavia, Italy
    Search for more papers by this author
  • Giuseppe Di Giulio M.D.,

    1. Istituto di Radiologia, Universitá di Pavia, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico S. Matteo, Pavia, Italy
    Search for more papers by this author
  • Monica Mantelli M.D.,

    1. Istituto di Radiologia, Universitá di Pavia, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico S. Matteo, Pavia, Italy
    Search for more papers by this author
  • Paola Anselmo M.D.,

    1. Istituto di Ematologia, Universitá “La Sapienza,” Roma, Italy
    Search for more papers by this author
  • Francesco Merli M.D.,

    1. Divisione di Ematologia, Arcispedale “S. Maria Nuova,” Reggio Emilia, Italy
    Search for more papers by this author
  • Pier L. Zinzani M.D.,

    1. Istituto di Ematologia “L e A. Seragnoli,” Universitá di Bologna, Policlinico S. Orsola-Malpighi, Bologna, Italy
    Search for more papers by this author
  • Gabriele Rossi M.D.,

    1. Istituto di Pediatria, Universitá di Pavia, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico S. Matteo, Pavia, Italy
    Search for more papers by this author
  • Vincenzo Callea M.D.,

    1. Divisione di Ematologia, Presidio Ospedali Riuniti, Reggio Calabria, Italy
    Search for more papers by this author
  • Emilio Iannitto M.D.,

    1. Cattedra e Divisione di Ematologia con Trapianto di Midollo Osseo, Universitá di Palermo, Policlinico, Palermo, Italy
    Search for more papers by this author
  • Marco Paulli M.D.,

    1. Istituto di Anatomia Patologica, Universitá di Pavia, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico S. Matteo, Pavia, Italy
    Search for more papers by this author
  • Lorena Garioni M.D.,

    1. Istituto di Radiologia, Universitá di Pavia, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico S. Matteo, Pavia, Italy
    Search for more papers by this author
  • Edoardo Ascari M.D.

    1. Medicina Interna e Oncologia Medica, Universitá di Pavia, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico S. Matteo, Pavia, Italy
    Search for more papers by this author

Abstract

BACKGROUND

The authors investigated the clinical role of tumor burden (TB) in patients with Hodgkin lymphoma, relating this parameter to most of the current clinical and prognostic factors and to the best predictive multifactorial models.

METHODS

The volume of TB at diagnosis was measured directly from the initial staging computed tomography scans in 351 patients who were treated on standard protocols. The mean patient age was 34.0 years ± 16.4 years. Forty-six patients had clinical Stage I disease, 201 patients had Stage II disease, 64 patients had Stage III disease, and 40 patients had Stage IV disease. There were 146 symptomatic patients. Overall survival (OS), disease-free survival (DFS), and time to treatment failure (TTF) were the time parameters evaluated in the multivariate analysis. Logistic regression was applied according to those who achieved or failed complete remission.

RESULTS

The mean TB normalized to body surface area (rTB) was 137.8 cm3/m2 ± 124.7 cm3/m2 (range, 1.9–694.5 cm3/m2). In multivariate analysis, rTB was the best predictor of TTF, DFS, and complete remission; the second best predictor of OS after patient age; and largely superior to all prognostic models analyzed. For the same stage and treatment, patients who were destined to clinical failure had an initial rTB 60–108% higher compared with the initial rTB in patients who achieved a cure, whereas differences in drug dose intensity were not significant.

CONCLUSIONS

In the current study, it was found that the rTB, as a prognostic factor, was more effective than and was independent of hitherto used factors and scores. The rTB may be a tool for evaluating the curative potential of treatment combinations, allowing physicians and patients to make better therapeutic choices earlier. Cancer 2004. © 2004 American Cancer Society.

Currently, approximately 70% of patients with Hodgkin disease can be cured definitively with the available treatments. How to free the remaining 30% of patients from lymphoma is a challenge, and the first problem is early and correct identification of these patients with unfavorable outcomes at the time of diagnosis. Various prognostic factors are used currently to differentiate treatments for patients with Hodgkin lymphoma (HL)1 according to the severity of clinical presentation. Such factors include gender, age, clinical stage, B symptoms, presence of bulky mass, serum albumin (Alb), hemoglobin (Hb), blood leukocyte count, and lymphocyte count.2, 3 However, although these factors generally have a good correlation with the clinical course of the disease, they all tend to have poor predictive ability. In fact, current methods of risk stratification, based on single factors or on prognostic models constructed with several selected parameters, are unable to identify a satisfactory proportion of truly high-risk patients. Currently, it seems reasonable that the patients who may benefit from intensified or investigational treatments can be defined as those with a < 50% probability of being free of disease progression at 5 years.4 To identify such patients, we need new and more reliable prognostic factors than those available currently. Tumor burden (TB), the importance of which was demonstrated clearly by Specht and colleagues 15 years ago,5–7 is an interesting and powerful prognostic factor. However, the original technique of evaluation (based on physical examination, chest X-ray, and bipedal lymphangiography) was semiquantitative, rather complex, and difficult to reproduce, and it never was applied subsequently by other investigators. Recently, some of us8, 9 demonstrated the feasibility of a more accurate measurement of TB through the evaluation of standard computed tomography (CT) images, confirmed the pivotal prognostic role of TB compared with all of the other clinical parameters currently used, and opened new prospects regarding early treatment choices.

Here, we present the results of TB evaluation in a large series. The retrospective nature of the study must be accepted in this first step, because it allowed quick collection of information from a sufficiently long follow-up to provide reliable indications of the true prognostic power of TB.

MATERIALS AND METHODS

Study Population and Management

Initial staging CT scans of the thorax, abdomen, and pelvis from 351 patients at 8 centers were reevaluated retrospectively to measure the volumes of all lymphomatous lesions that were detectable before the start of treatment. In addition to the availability of CT magnetic records or radiographs, patients had to fulfill the following inclusion criteria: 1) biopsy-proven HL; 2) first diagnosis, staging, and treatment after 1989; 3) availability of ultrasonographic measurements (with at least 2 measures of greatest dimension) of the lymph nodes outside thoracic and abdominal CT scans, if any; 4) at least unilateral bone marrow biopsy and basic evaluation of differential blood count, erythrocyte sedimentation rate (ESR), Hb, Alb, and lactate dehydrogenase (LDH); 5) staging categorization, records of systemic symptoms, bulky mass evaluation, and grading of therapeutic outcome as complete response (CR), partial response (PR), null response (NR), and progressive disease (PD) performed according to the Cotswolds meeting criteria2; 6) evaluation of performance status by means of the Karnofsky index; and 7) therapy performed according to standard protocols within well known, controlled or randomized trials.

Serum concentration of β2-microglobulin was not considered absolutely necessary and was available in 110 patients. Sites of disease involvement basically were considered according to the diagram of the anatomic definition of separate lymph node regions proposed by Kaplan and Rosenberg10 with a few modifications. In particular, the large radiotherapeutic area of the cervical site was split into two distinct regions: the strictly cervical region (in the proper anatomic sense) and the supraclavicular region (as described by Specht7). Moreover, the infraclavicular region was considered together with the axillary region, in accordance with Vassilakopoulos et al.11 Thus, the following supradiaphragmatic regions were counted: Waldeyer ring, cervical, supraclavicular, axillary and/or infraclavicular, mediastinal, hilar, and epitrochlear. The subdiaphragmatic regions considered were paraaortic, spleen (with hilar lymph nodes), mesenteric, iliac, inguinal and/or femoral, and popliteal. Each extranodal site was counted separately, contributing to the total number of involved sites (NIS).

The histology of diagnostic specimens was reviewed. Patients with any stage of disease were accepted in the study. Therapy differed as follows: Seven patients with Stage I disease who had a very favorable clinical presentation underwent extended-field radiotherapy alone; 53 patients with Stage I and II disease who had no more than 1 factor present among extranodal lesion, ESR >40 mm at the first hour, and LDH higher than normal level, received chemotherapy with vinblastine, bleomycin, and methotrexate (VBM)12 combined with extended-field radiotherapy (before 1997) or involved-field radiotherapy (after 1997); 126 patients with Stage I and II disease who had an unfavorable presentation because of lymphocyte-depleted histologic type or bulky mass or who had more than 1 factor among extranodal lesion, high ESR, and high LDH, were treated with 4 cycles of chemotherapy13 with doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD) followed by extended-field radiotherapy; 172 patients with Stage IIB, III, or IV disease were treated with 6 cycles of ABVD, or with alternating mechloretamine, vincristine, procarbazine, and prednisone (MOPP)/ABVD,14 or with a hybrid schedule providing mechloretamine, vincristine, procarbazine, prednisone, epidoxorubicin, bleomycin, vinblastine, lomustine, melphalan, and vindesine (MOPPEBVCAD).15 In these patients with advanced-stage disease, radiotherapy was added optionally to the chemotherapy and was delivered only to lymphomatous lesions that responded slowly during chemotherapy or incompletely after the end of chemotherapy.

Details on drug doses and time of administration of chemotherapy were available for 229 of 351 patients, and dose intensities of both single drugs and whole regimens were calculated according to the method of Hryniuk and Bush.16 The current series included 121 patients in whom the feasibility of TB measurement through CT scans was evaluated and reported first.8 The follow-up of these patients was prolonged by an average of 2.5 years; 3 additional recurrences and 1 cardiac sudden death occurred among the patients in this period. The median follow-up of the whole study population was 62 months (range, 9–198 months).

TB Assessment

The technical procedures for evaluating CT scans to measure TB have been detailed carefully elsewhere.8 In brief, scans had to be no more than 10 mm thick and contiguous. They could be generated either by a spiral unit or by a conventional unit; for nonspiral CT units, the scanning time had to be no longer than 5 seconds. Images had to be taken with nonionic contrast medium within an angio-CT scan evaluation of the upper abdominal parenchymal organs. The majority of measurements were performed through a reevaluation of images saved on magnetic records; however, in some patients, the images had to be digitalized again from the radiographic films (a series of double measurements with the two methods showed equivalent accuracy). Radiologists, who were blind to all other clinical information, systematically outlined every lymphomatous lesion recognizable on each scan slice. This allowed the software resources of the CT scanner to calculate the areas of the lesions and, from the thickness of the slice, the volume of the amount of tumor per slice and, finally, the sum of the volumes of all the slices. Healthy structures surrounded by or included in lymphomatous tissue were subtracted, and necrotic areas were included. Visceral lesions were handled like lymph node lesions. The calculation procedures took the radiologists 20–30 minutes per patient. Both the variability and the reproducibility of TB measurements were checked for 22 patients whose CT scans were assessed separately and blindly by the 3 radiologists involved in this study (G.D.G., M.M., and L.G.). The means ± standard deviations (with variability ranges) of the percentage errors recorded were as follows: − 0.8% ± 5.4% (range, from − 6.6% to + 8.3%) for the results of M.M. versus the results of G.D.G.; − 0.3% ± 4.6% (range, from − 9.4% to + 6.3%) for the results of L.G. versus the results of G.D.G.; and + 0.3% ± 5.5% (range, from − 7.8% to + 8.7%) for the results of L.G. versus the results of M.M. The t test for paired data demonstrated that the 3 series of results substantially were similar (M.M. vs. G.D.G.: t = 0.084; P = 0.9342; L.G. vs. G.D.G.: t = 0.591; P = 0.6081; L.G. vs. M.M.: t = 0.626; P = 0.5382).

Because bone marrow involvement escapes the CT technique, to incorporate this into the measurement of TB, first, we had to derive the volume of the hematopoietically active bone marrow from the formula of Wickramasinghe17 (hematopoietic bone marrow volume = 20 mL/Kg body weight). To minimize the possible errors induced by the variable proportion of fatty mass, we chose to apply this formula to the ideal body weight (IBW) rather than the patient's actual weight. IBW was derived from the following equation proposed by Devine18: IBW = 50 Kg (for males) or 45.5 Kg (for females) + 2.3 Kg per inch (2.54 cm) of height > 5 feet (152.4 cm). The bone marrow TB was then calculated as 1/3 of the calculated volume when involvement was diffuse, 1/10 when it was focal, and 1/20 when it was defined as nodular.8

The few patients who had involved superficial lymph nodes outside the CT scans were considered in this study only if they had ultrasonographic measurements of at least two greatest dimensions of each involved lymph node. The volumes of these superficial lesions were calculated geometrically and then summed with those from the CT evaluation to reach the total TB. This was expressed in cubic centimeters and normalized to squared meter of body surface (rTB).

Statistics

Differences in rTB values related to the main clinical characteristics of the patients were analyzed with the t test for unpaired data.19 Survival was calculated with the Kaplan–Meier method,20 and differences between curves were evaluated statistically with the log-rank test.21 Multiple regression analysis19 was used to explore the predictability of rTB from common clinical parameters and to evaluate the relation between pretherapeutic clinical factors and both the achievement of CR and the ability to maintain this status. Simple and multiple regression analyses were applied to the proportional hazards model22 to explore the prognostic role of rTB compared with the roles of other clinical parameters. The dependent variables for these tests were the time to treatment failure (TTF) (from the start of treatment to disease progression during treatment or less than CR at the end of it, or recurrence, or death from disease at any time), disease-free survival (DFS) (from the end of treatment, for patients who achieved a CR, to recurrence or death from disease), and overall survival (OS) (from diagnosis to death from any cause).23 The same technique, related to TTF, was used to compare the predictive value of rTB with that of the best prognostic models devised for HL, i.e., the model from the International Database on Hodgkin's Disease (IDHD),24 the model produced at the Memorial Sloan-Kettering Cancer Center (MSK),25 the score of the International Prognostic Factor Project on Advanced Hodgkin's Disease (IPFP),3 and the index recently created from an integration of all three models (Integrated Index26) after the demonstration of their superiority over others.

RESULTS

The mean value of the rTB recorded for the entire study population was 137.8 cm3/m2 ± 123.9 cm3/m2, with limits ranging from 1.9 cm3/m2 to 694.5 cm3/m2. Figure 1 shows the percentile distribution of rTB data, illustrating that the median value was 102.1 cm3/m2. Tables 1 and 2 show the differences in rTB values related to the main clinical characteristics and laboratory parameters found at the initial staging of the disease. Males had greater tumor masses than females, although the difference was not statistically significant, and there was a trend toward an inverse correlation between rTB and age at diagnosis. Overall, greater tumor volumes appeared to be related clearly to advanced disease stage, systemic symptoms, bulky mass, lymphocyte-depleted histology (although very few patients had this histologic type), higher NIS, lower concentrations of Hb and Alb, and higher levels of ESR, LDH and β2-microglobulin. Nearly all groups of patients identified by these staging parameters could be split further into two subgroups, each with very different prognoses, according to whether the patients had a rTB values greater than or less than the mean value for that group. Figures 2–5 demonstrate how strongly rTB can improve the prognostic discrimination provided by four of the main parameters currently utilized to differentiate available treatment strategies, i.e., stage, including both early (Fig. 2) and advanced (Fig. 3) disease stages; the absence or presence of systemic symptoms (Fig. 4, Groups A and B); and the absence or presence of bulky mass (Fig. 5). Figure 6 shows the remarkable prognostic improvement provided by the evaluation of rTB after Alb evaluation as a representative example of what happens with all of the laboratory parameters considered in Table 2.

Figure 1.

Percentile distribution of the mean tumor burden normalized to body surface area (rTB) data in the 351 patients. The box reports the rTB values corresponding to the 10th, 25th, 50th, 75th, and 90th percentiles.

Table 1. Values of the Mean Tumor Burden Normalized to Body Surface Area According to the Main Clinical Staging Characteristics
CharacteristicNo. of patientsMean rTB ± SD (cm3/m2)P value (t test)
  1. rTB: mean tumor burden normalized to body surface area; SD: standard deviation; NS: not significant; LP: lymphocyte-predominant; NS: nodular sclerosis; MC: mixed cellularity; LD: lymphocyte-depleted; Incl.: inclassifiable.

  2. aP < 0.050 for the following comparisons: age groups (age < 15 yrs) vs. (ages 45–59 yrs) and (age ≥ 60 yrs) vs. (every other age group except ages 45–59 yrs), clinical stage (Stage I vs. every other stage and Stage II vs. Stage IV), and histologic type (lymphocyte-depleted vs. every other type).

Gender
 Male168145.4 ± 124.7NS
 Female183130.8 ± 123.0NS
Age (yrs)
 < 1523178.6 ± 169.1NS
 15–29151149.1 ± 117.5NS
 30–4494140.0 ± 126.8NS
 45–5949119.4 ± 127.8NS
 ≥ 603480.4 ± 79.1 
Clinical stage
 Stage I4687.6 ± 85.60.039
 Stage II201127.6 ± 113.5NS
 Stage III64153.5 ± 134.20.012
 Stage IV40221.8 ± 142.0 
Systemic symptoms
 A205107.4 ± 102.65.6 × 10−8
 B146180.5 ± 138.2 
Histologic type
 LP17104.8 ± 101.5NS
 NS245136.8 ± 119.8NS
 MC73139.0 ± 123.30.015
 LD10240.7 ± 200.60.017
 Incl.687.7 ± 67.00.017
No. of involved sites
 15090.7 ± 87.1NS
 2104107.8 ± 92.4NS
 370133.8 ± 119.4NS
 450169.3 ± 152.5NS
 536194.4 ± 143.8NS
 615159.6 ± 80.8NS
 714202.5 ± 110.2NS
 83170.3 ± 107.8NS
 93264.8 ± 60.7NS
 ≥ 103192.8 ± 105.7 
Bulky mass
 Absent24698.9 ± 95.81.5 × 10−20
 Present105225.0 ± 132.6 
Table 2. Values of the Mean Tumor Burden Normalized to Body Surface Area According to the Levels at Diagnosis of the Prognostic Laboratory Tests Currently Used for Patients with Hodgkin Lymphoma
Laboratory testNo. of patientsMean rTB ± SD (cm3/m2)P value (t test)
  • rTB: mean tumor burden normalized to body surface area; SD: standard deviation; NS: not significant.

  • a

    Values were missing for 111 patients.

  • b

    Values were missing for 241 patients.

Hemoglobin (g/dL)
 ≥ 10.5295121.8 ± 106.7 
 < 10.556216.1 ± 165.71.2 × 10−7
Erythrocyte sedimentation rate (mm/first hr)
 < 4016994.4 ± 91.8 
 ≥ 40182175.5 ± 131.08.2 × 10−10
Fibrinogen (mg/mL)a
 ≤ 450142108.5 ± 110.3 
 > 45098163.9 ± 129.10.0004
Albumin (g/dL)
 ≥ 3.5255118.6 ± 112.9 
 < 3.596187.5 ± 130.62.7 × 10−6
Lactate dehydrogenase (U/L)
 ≤ 460286119.5 ± 111.4 
 > 46065217.8 ± 141.82.3 × 10−8
β-2-microglobulin (μg/L)b
 ≤ 300095127.1 ± 115.6 
 > 300015219.9 ± 139.00.0060
Figure 2.

Time to treatment failure (TTF) curves for patients with Stage I and II disease were calculated according to whether their mean tumor burden normalized to body surface area (rTB) was ≤ or > the mean value for each stage (≤ 87.6 cm3/m2 or > 87.6 cm3/m2 for patients with Stage I disease; ≤ 127.6 cm3/m2 or > 127.6 cm3/m2 for patients with Stage II disease).

Figure 3.

Time to treatment failure (TTF) curves for patients with Stage III and IV disease were calculated according to whether the mean tumor burden normalized to body surface area (rTB) was ≤ or > the mean value for each stage (≤ 153.5 cm3/m2 or > 153.5 cm3/m2 for patients with Stage III disease; ≤ 221.8 cm3/m2 or > 221.8 cm3/m2 for patients with Stage IV disease).

Figure 4.

Time to treatment failure (TTF) curves for patients without (Group A) and with (Group B) general symptoms were calculated according to whether their mean tumor burden normalized to body surface area (rTB) was ≤ or > the mean value for each group (≤ 107.4 cm3/m2 or > 107.4 cm3/m2 for patients in Group A; ≤ 180.5 cm3/m2 or > 180.5 cm3/m2 for patients in Group B).

Figure 5.

Time to treatment failure (TTF) curves for patients with (+) or without (−) bulky mass were calculated according to whether their mean tumor burden normalized to body surface area (rTB) was ≤ or > the mean value for each group (≤ 98.9 cm3/m2 or > 98.9 cm3/m2 for patients without bulk; ≤ 225.0 cm3/m2or > 225.0 for patients with bulk).

Figure 6.

Time to treatment failure (TTF) curves for patients with high or low concentrations of serum albumin (Alb) (≥ 3.5 g/dL or < 3.5 g/dL, respectively) were calculated according to whether the mean tumor burden normalized to body surface area (rTB) was ≤ or > the mean values for each group (≤ 118.6 cm3/m2 or > 118.6 cm3/m2 for patients with high Alb levels; ≤ 187.5 cm3/m2 or > 187.5 cm3/m2 for patients with low Alb levels).

However, the primary prognostic role of rTB is demonstrated by the results presented in Tables 3 and 4. Table 3 shows that rTB was the strongest predictive factor for TTF, DFS, and clinical outcome within the first 12 months and the second strongest factor, after age, for OS (it is well known that age is the first prognosticator for survival in the majority of adult diseases). Table 3 demonstrates that rTB predicts TTF better than the three best prognostic models devised for HL—IDHD, MSK, and IPFP—in addition to the composite Integrated Index, which should best integrate these three models. Overall, the prognostic importance of rTB overwhelms that of most of the clinical factors currently used. Moreover, the subordinate role played by the NIS in three of four analyses of Table 3 seems to suggest that the distribution of TB is the second main, and often the only, statistically predictive factor after the absolute amount of TB.

Table 3. Prognostic Value of the Mean Tumor Burden Normalized to Body Surface Area Emerging from Multivariate Analysis with Different Dependent Variables: Proportional Hazards Model for Time to Treatment Failure, Disease-Free Survival, and Overall Survival with Logistic Multiple Regression for the Variable “Less than Complete Response or Early Recurrence versus Otherwise”
Dependent variableStepwise selection of best covariatesCoefficientP valueVariables not entered in the model (i.e., with P > 0.05), listed in increasing order of their P values
  1. TTF: time to disease failure; rTB: mean tumor burden normalized to body surface area; NIS: number of involved areas; ESR: erythrocyte sedimentation rate, Alb: serum albumin; Hb: hemoglobin; LDH: serum lactate dehydrogenase; DFS: disease-free survival; OS: overall survival; CR: complete response.

TTFrTB0.0055.8 × 10−9Disease stage, bulky mass, performance status, age, ESR, Alb, Hb, gender, symptoms, LDH, histology, bone marrow involvement
NIS0.1880.0002
DFSrTB0.0015.3 × 10−6Performance status, ESR, bulky mass, Hb, disease stage, Alb, symptoms, gender, LDH, bone marrow involvement, age
NIS0.0630.0011
Histology0.1830.0388
OSAge0.0456.2 × 10−5NIS, gender, Hb, ESR, bulky mass, Alb, LDH, histology, symptoms, bone marrow involvement
rTB0.0050.0005
Disease stage0.5150.0139
Performance status−0.0270.0428
< CR or early recurrence/otherwiserTB0.0013.5 × 10−10ESR, LDH, stage, symptoms, Alb, histology, bulky mass, bone marrow involvement, performance status, Hb
Age0.0030.0156
NIS0.0220.0209
Table 4. Prognostic Value of the Mean Tumor Burden Normalized to Body Surface Area in Relation to the Time to Treatment Failure and Compared with the Value of the Best Multiparametric Prognostic Indices
All 351 patients172 Patients with advance-stage disease (Stage IIB, III, IV)
VariableCoefficientP valueVariableCoefficientP value
  1. rTB: mean tumor burden normalized to body surface area; IDHD: International Database on Hodgkin's Disease (for the log-normal model of the IDHD, see Gobbi et al., 199424); MSK: Memorial Sloan-Kettering Cancer Center (for the scoring system proposed by MSK, see Straus et al., 199025); IPFP: International Prognostic Factor Projection on Hodgkin's Disease (for the scoring system proposed by the IPFP, see Hasenclever and Diehl, 19983); Int. Ind.: Integrated Index (composite of the independent information for all three of the preceding indices; see Gobbi et al., 200126).

rTB0.0063.3 × 10−15rTB0.0051.8 × 10−8
IDHD−2.3853.2 × 10−6IDHD−2.0340.0006
MSK−0.3660.0196MSK3.8090.0510
Int. Ind.0.2240.6360IPFP1.1440.2847
IPFP0.1180.7360Int.Ind.0.4340.5101

Almost all of the clinical and staging parameters are correlated partially with rTB, but they cannot be used to estimate rTB with enough accuracy. In fact, the best selection of parameters (rTB = 305.2 + 106 × bulky mass [0/1] − 18.8 × Hb [g/dL] + 11.2 × NIS [absolute number] + 0.1 × LDH [mU/mL] − 30.9 + gender [1/2]) showed a low R2 (0.400) and very low predictive power, with an unacceptable mean error of + 124.3% ± 449.1%. The same regression performed with β2-microglobulin included in the model (but with only 110 patients) showed similar results (rTB = 211.6 + 97.9 × bulky mass [1/0] + 14.4 × NIS [absolute number] − 14.7 × Hb [g/dL] + 0.02 × β2-microglobulin [μg/mL]; R2 = 0.428; mean error = + 79.0% + 154.8%). Thus, rTB cannot be deduced from the current clinical parameters and must be measured directly.

The rising risk of treatment failure related to increasing amounts of initial TB is shown in Table 5, in which the main parameters of clinical outcome are reported by rTB quartiles. Recurrence rates are higher, and both DFS and TTF at 10 years are lower in patients with a high rTB. Complete remission rates and OS rates were similar in the first three rTB quartiles, with a sharp fall only in the fourth quartile. This may have been due to some inaccuracy in disease evaluation (unrecognized minimal residual disease, erroneously classified as complete remission, may be classified later as an early recurrence) and, with regard to survival, may have been due to the effectiveness of salvage therapies (probably less effective only in the presence of high TB).

Table 5. Complete Remission Rates, Recurrence Rates, and 10-Year Rates of Disease-Free Survival, Time to Treatment Failure, and Overall Survival Recorded in the Study Population According to Quartiles of the Mean Tumor Burden Normalized to Body Surface Area
CharacteristicQuartile
FirstSecondThirdFourth
  1. rTB: mean tumor burden normalized to body surface area; CR: complete response; DFS: disease-free survival; TTF: time to treatment failure; OS: overall survival.

rTB (cm3/m2)< 41.041.0–101.9102.0–191.9≥ 192.0
No. of patients87898788
CR rate0.940.950.940.82
Recurrence rate0.050.120.160.30
10-Yr DFS0.91 ± 0.040.83 ± 0.050.78 ± 0.070.57 ± 0.06
10-Yr TTF0.88 ± 0.040.83 ± 0.050.75 ± 0.070.49 ± 0.06
10-Yr OS0.87 ± 0.050.92 ± 0.040.97 ± 0.030.67 ± 0.08

To investigate the clinical role of rTB further, we focused on the 229 patients whose complete details regarding chemotherapy and radiotherapy were available. We tested for possible differences between patients who responded completely and did not develop recurrent disease and patients who had less than a CR, or developed recurrent disease at any time, or died of disease (i.e., TTF events). Within each treatment protocol adopted, there were no statistically significant differences in drug dose intensities or radiotherapy doses and fields between patients who had successful treatment and patients who had unsuccessful treatment. However, within each therapeutic group, patients who had treatment failure had much higher starting rTB volumes, with a dramatic (and statistically significant) difference compared with patients who responded completely and permanently. Table 6 reports the results of this analysis related to 37 patients with early-stage disease who had a favorable prognosis and were treated with VBM chemotherapy, 79 patients with early-stage disease who had an unfavorable prognosis and were treated with the ABVD regimen, and 113 patients with advanced-stage disease who were treated with the ABVD schedule (n = 56 patients) or with alternating or hybrid MOPP/ABVD or cyclophosphamide, vincristine, procarbazine, and prednisone/ABVD chemotherapy (n = 57 patients). The comparison presented in Table 6 shows that the differences in average regimen dose intensity between patients who failed treatment and patients who were treated successfully were minimal (1–2%), whereas the differences in rTB ranged from 60% to 108%. There were no differences in the number of chemotherapy cycles administered, no marked deviations from the average of each regimen in single-drug dose intensities, and no differences in the radiotherapy doses employed. In only one group was a significant difference found other than rTB: In patients with advanced disease who were treated with alternated eight-drug schedules, the NIS was significantly greater in patients who failed treatment compared with the number of sites in patients who were treated successfully. Despite the relatively low number of patients in whom drug doses were analyzed, the discrepancy between the large differences in rTB and the minimal variations of treatment-related parameters may offer keys to interpreting the clinical failures.

Table 6. Differences in the Mean Tumor Burden Normalized to Body Surface Area Values Found at Diagnosis among Patients Destined to Fail and Patients who Were Cured within Distinct Groups of Homogeneously Treated Patientsa
Patient groupMean ± SDP value
No failuresFailures
  • SD: standard deviation; VBM: vinblastine, bleomycin, and methotrexate; IF-RT: involved-field radiotherapy; rTB: mean tumor burden normalized to body surface area; TDI: total dose intensity; NS: not significant; NIS: number of involved sites; ABVD: doxorubicin, bleomycin, vinblastine, and dacarbazine; EF-RT: extended-field radiotherapy; M(C)OPP: mechloretamine (or cyclophosphamide), vincristine, procarbazine, and prednisone; RT: radiotherapy. TDI: proportion of planned doses and time and actual doses and time.

  • a

    Data on average dose intensity of the whole chemotherapy regimens (TDI) and the mean NIS are reported for each group.

Stage IA–IIA, favorable (VBM plus IF-RT)
 No. of patients325 
 rTB (cm3/m2)53.6 ± 42.7111.5 ± 76.40.0164
 TDI0.94 ± 0.181.15 ± 0.31NS
 NIS (no.)2.6 ± 1.43.0 ± 1.2NS
Stage IA–IIA, unfavorable (ABVD × 4 plus EF-RT)
 No. of patients727 
 RTB (cm3/m2)1110.5 ± 86.2177.3 ± 89.90.0320
 TDI0.87 ± 0.120.88 ± 0.09NS
 NIS (no.)2.3 ± 1.22.7 ± 1.4NS
Stages IIB, III–IV (ABVD × 6–8 ± RT)
 No. of patients4016 
 RTB (cm3/m2)161.3 ± 99.8279.0 ± 145.20.0010
 TDI0.88 ± 0.150.93 ± 0.13NS
 NIS (no.)3.7 ± 1.74.2 ± 1.8NS
Stage IIB, III–IV (M[C]OPP/ABVD or equivalents × 6 ± RT)
 No. of patients4512 
 rTB (cm3/m2)143.7 ± 103.1235.8 ± 142.20.0143
 TDI0.82 ± 0.190.89 ± 0.25NS
 NIS (no.)3.8 ± 1.86.0 ± 3.40.0401

DISCUSSION

The trend toward an inverse correlation between rTB and age at diagnosis may have been related to the well recognized, increasing prevalence in histologic types characterized by a relatively larger population of Hodgkin and Reed–Sternberg cells, a smaller inflammatory component in older patients,27 and the lower reactivity of immunologic cells in the elderly. Because TB is a measure of the total volume of tumor lesions, in which the reactive component generally predominates strongly over the neoplastic component, this may explain why the TB at diagnosis is progressively lower as the patient's age increases. Overall, the results of the current study demonstrate that rTB is the most important prognostic factor in HL. Its superiority emerges from direct comparisons with most of the currently used prognostic parameters in relation to several clinical endpoints. Moreover, rTB alone showed a leading role over the three best multifactorial models in HL and even over the complex index that integrated all three models. Most of the well known, elementary prognostic factors for HL showed a variable, generally marked intercorrelation with rTB and lost a large part of their predictive power when rTB was taken into account in the analyses. To demonstrate this, we intentionally avoided showing survival curves related to different ranges of rTB volumes, because every choice of optimal cut-off levels would reflect a search for best data fitting, whereas the results shown in Tables 3–5 derive from the analysis of the whole unmanipulated distribution of the data as they were found in the original clinical files or as they were given by radiologists (who were blinded to the clinical data). However, apart from the values of rTB quartiles employed for statistical requirements (Table 5), we can confirm that, for clinical purposes, the ranges suggested elsewhere8 for rTB < 90 cm3/m2, from 90–280 cm3/m2, and > 280 cm3/m2 still seem to be very interesting for discriminating patients with a low, intermediate, or high risk within any given survival parameter.

All of these achievements strengthen the original idea and the pioneer investigations of Specht,7 who used an indirect and semiquantitative method, and they confirm the preliminary results from the feasibility study done by some of us8 on this new, direct, quantitative technique for TB measurement. We demonstrated that this technique was equally accurate on images from the original CT radiographic films and on images from magnetic records (see Material and Methods). Thus, it can be applied retrospectively to either magnetic records or radiographic films of staging CT evaluation. In retrospective studies and CT scans that do not cover the entire body, the only additional requirement is the availability of ultrasonographic measurements of superficial lesions outside CT scans (if any). For this purpose, it has been demonstrated that even an accurate physical evaluation of two greatest dimensions offers a reliable volumetric assessment of high cervical and inguinal lymph nodes (although not of supraclavicular or axillary lymph nodes).28

The number of involved anatomic sites can be considered the second most important prognostic factor. We agree with Vassilakopoulos et al.11 about the ease and wide applicability of calculating the NIS, although it is evident both from the original results of Specht and colleagues and from the current study that, when rTB is taken into account, the prognostic significance of the NIS decreases strongly. Many other factors that are referred to traditionally as characteristics of tumor cells or of the host or parameters, depending on the interaction of both, and historically considered the foundations of therapeutic choices (disease stage, B symptoms, bulky mass, etc.) have greatly reduced significance or no further significance after a regression analysis with the variables rTB and NIS. These factors probably reflect only part of the prognostic significance of rTB and may be considered variably dependent on this latter.

Our data suggest that characteristics closely connected to the tumor itself, such as absolute amount and anatomic spread, are primary factors. This assessment may match the current evolution of HL treatment better. Thirty years ago, when radiation was considered the primary therapy for HL, and chemotherapy had an ancillary role, the staging system was devised as the logical consequence of the radiotherapists' need to know the exact limits of the tumor and its spread to plan radiotherapy correctly. The current staging system, which was revised at the Cotswolds meeting in 1989, is a slightly modified version of those proposed at the Rye and Ann Arbor conferences29, 30 and still reflects radiotherapy-oriented bases. In fact, we still consider sites of very different sizes in the list of anatomic areas of possible involvement, mainly because of their different radiotherapeutic requirements (e.g., the cervical site, which includes occipital, preauricular, cervical, and supraclavicular lymph nodes). Similarly, we still accept the diaphragm as a magic border separating Stage II and III disease. It is clear that, in the past, the dramatic clinical success of radiotherapy—when planned based on accurate knowledge of tumor extension—largely justified these working concepts up to the point of constructing a rationale for exploratory laparotomy with splenectomy to detect occult sites of abdominal disease. In general, exploratory laparotomy has now been abandoned because of the improvement in modern imaging techniques and the increasing role of chemotherapy in treatment programs. From this point of view, it was to be expected that the prognostic power of traditional anatomic stages, symptoms, and bulky masses, which were devised primarily for locally active therapy, would be overcome by a parameter that interacts more closely with systemically active treatments. Thus, the evaluation of rTB may contribute more to clinical investigations of the reasons for treatment failures, of the indications for optimal treatment, and of the actual possibility for cure offered by different chemotherapy regimens. For example, it would be very interesting to search for possible threshold volumes of rTB beyond which a given regimen has too high a risk of failure. Although the number of patients considered is too low to make firm statements, a working hypothesis—drawn from the means and standard deviations reported in Table 6 and to be validated in future investigations—may be to consider 70–80 cm3/m2 as the highest rTB for low-impact treatments, such as 6 cycles of VBM plus involved-field radiotherapy; 120–130 cm3/m2 for 4 cycles of ABVD cycles plus extended-field radiotherapy; and 180–190 cm3/m2 for 6–8 cycles of ABVD or intense 6–8-drug, alternated or hybrid regimens. Similar limits of rTB, when they have been validated widely, may provide a criterion for indicating early high-dose therapies or other investigational treatments. A multicenter, prospective study by the Italian Lymphoma Intergroup is starting to explore this matter more thoroughly.

Ancillary