A comprehensive geriatric assessment is more effective than clinical judgment to identify elderly diffuse large cell lymphoma patients who benefit from aggressive therapy




The authors set out to analyze if a simple comprehensive geriatric assessment (CGA) could objectively identify elderly patients with diffuse large cell lymphoma (DLCL) who can be effectively treated with anthracycline-containing immunochemotherapy.


CGA was performed in 84 consecutive patients with DLCL aged >65 years and diagnosed at a single institution. Treatment with curative versus palliative intent was chosen according to clinical judgment. Cyclophosphamide, hydroxydaunomycin, Oncovin (vincristine), and prednisone (CHOP) or CHOP-like regimens were given to 62 (74%) patients. The outcome of patients was analyzed according to both the treatment received and the results of CGA.


According to CGA, 42 (50%) patients were classified as “fit.” They were younger (P < .0001) and had less frequent systemic symptoms (P = .03). These patients received curative treatment by clinical judgment. Their response rate (92.5% vs 48.8%; P < .0001) and median survival (not reached vs 8 months; P < .0001) were significantly better than those of 42 patients considered “unfit” by CGA. Among unfit patients, 20 had actually received curative and 22 palliative therapy. These subgroups did not differ in any geriatric or lymphoma-related characteristic. Their outcome was similar irrespectively of the type of treatment received (median survival, 8 vs 7 months; P = nonsignificant). Lymphoma rather than toxicity was the main cause of failure/death also among unfit patients treated aggressively.


CGA is an efficient method to identify elderly DLCL patients who can benefit from a curative approach with anthracycline-containing immunochemotherapy. Further study is needed to discern why unfit patients seem to have poor outcomes because of poor tolerance but also because of lymphoma refractoriness to intensive therapy. Cancer 2009. © 2009 American Cancer Society.

Diffuse large cell lymphoma (DLCL) is the most frequent type of malignant lymphoma. Its incidence is increasing,1 as is the cure rate with a combination of aggressive chemotherapy and monoclonal antibodies. The median age of DLCL patients at presentation is 65 years. Therefore, it is 1 of the more frequent among neoplastic diseases of the elderly, in which cure can be achieved when optimal medical treatment can be safely delivered. Clinical presentation and prognostic features are similar between young and elderly patients with DLCL, but the outcome is worse among elderly patients, particularly because of comorbidities, diminished organ function, and impaired drug metabolism characteristic of aging.

Although several low-dose chemotherapy regimens, tailored for elderly patients, have been proposed, recent studies have shown that survival of older patients with DLCL is best improved by the use of a standard chemotherapy regimen, like cyclophosphamide, hydroxydaunomycin, Oncovin (vincristine), and prednisone (CHOP), at full doses,2, 3 with the addition of rituximab.4 However, because using aggressive therapies in elderly patients carries a high risk of severe toxicity and death,5 the major challenge for clinicians treating unselected elderly patients with DLCL is to identify, as accurately as possible, those who can safely tolerate full-dose treatment.

There are no validated methods to prospectively identify elderly patients fit enough to receive the same treatment as younger patients. Recently, comprehensive geriatric assessment (CGA) has been proposed as an objective tool for supporting medical decision making, but its efficacy has yet to be formally demonstrated.6, 7

In the present study, a CGA has been performed in consecutive elderly patients with DLCL, whose treatment was selected according to clinical evaluation, to objectively evaluate the potential usefulness of CGA categorization as a predictor of treatment tolerability and outcome.


In all consecutive patients aged >65 years presenting with newly diagnosed DLCL, a CGA was performed in addition to staging procedures. CGA included assessment of the following 4 parameters: 1) age ≥80 years; 2) activity of daily living (ADL), that is, loss in any activity, including bathing, dressing, toileting, transferring, feeding, and continence8; 3) comorbidity score according to the Cumulative Illness Rating Score for Geriatrics (CIRS-G) and evaluated in all the organs/systems9 as follows: no problem—1, mild problem (may require treatment)—2, moderate disability or morbidity (treatment required)—3, severe, constant, significant disability/”uncontrollable” chronic problems—4, extremely severe disability, immediate treatment required/end organ failure/severe impairment in function—5; and 4) geriatric syndrome, defined as the occurrence of dementia, delirium, depression, incontinence, falls, osteoporosis, neglect and abuse, or failure to thrive, as detailed in Balducci and Beghe.6 After written informed consent, patients were classified in the category of “fit” patients if they were aged <80, had an ADL score of 6, <3 grade 3 CIRS-G comorbidities and no grade 4 comorbidities (hematological comorbidities were not investigated), and none of the criteria defining the presence of geriatric syndrome. All other patients were classified as “unfit.”

The decision to treat a patient and the choice of the type and intensity of treatment were always left to the clinical judgment of the attending physician, who was blind to the result of CGA. Full-dose treatment with curative intent was considered the use of a combination of an anthracycline-based chemotherapy, consisting of CHOP or CHOP-like regimens, with rituximab. Patients considered unable to tolerate such treatment received palliative therapy, including radiation therapy, low-dose chemotherapy without anthracyclines (cyclophosphamide, Oncovin [vincristine], and prednisone [COP], low-dose COP), rituximab as a single agent, and corticosteroids.

Statistical analyses were performed using Prism software (GraphPad Software, La Jolla, Calif). The characteristics of the subgroups of patients subdivided according to the type of treatment received and to CGA categorization were compared using Fisher exact test, Student t test, and Mann-Whitney statistics, as appropriate. Log-rank analysis was used to compare actuarial survival curves.


From January 2003 to December 2006, 88 patients aged >65 years (61% were aged ≥70 years) were consecutively diagnosed as affected by DLCL, and 84 of them had fully evaluable data and were considered for the present study. Their median age was 73 years (range, 66-89 years), 66% were in stage III-IV, 32% had B symptoms, and 63% belonged to the categories of intermediate-high or high risk according to the International Prognostic Index.10 On the basis of clinical judgment, 62 patients (74%) received full-dose therapy with curative intent, and 22 received palliation. The proportion of patients treated aggressively was identical to that recorded in the historical cohort of elderly DLCL patients diagnosed and managed between 1995 and 2002 at our institution, when 64 of 86 (74.4%) consecutive patients received aggressive treatment.

The response rate (RR) (complete remission + partial remission) of patients treated aggressively was 79.7% (47 of 59), compared with 50% (11 of 22) in patients receiving palliation (Fisher exact test; P = .013). The 2-year actuarial progression-free survival (PFS) and overall survival (OS) were significantly better among patients treated with curative intent than among patients receiving palliation (PFS: 55.9% vs 22.2%; log-rank analysis: P = .0002) (OS: 57.7% vs 26.1%; log rank analysis: P = .0014) (Fig. 1).

Figure 1.

Actuarial overall survival curves of elderly patients with diffuse large cell lymphoma given intensive treatment or palliation according to physician's clinical evaluation are shown (log-rank analysis: P = .0014).

According to CGA, 42 of 84 (50%) patients were classified as fit and 42 (50%) as unfit. The criteria for classifying a patient as unfit were age in 15 (35.7%), low ADL in 4 (11.9%), comorbidity in 20 (50%), and geriatric syndrome in 3 cases (7.1%), respectively (Table 1). All the patients with geriatric syndrome had further concomitant reasons for being considered unfit (low ADL score in 1, comorbidity in 1, both low ADL and comorbidity in 1). The 2 subgroups of fit and unfit patients differed significantly in mean age (70.8 years vs 76.3 years; P < .0001).

Table 1. Distribution of the Causes of Unfitness According to the Different CGA Parameters Considered Among the Subgroups of Unfit Patients Receiving Treatment With Curative or Palliative Intent
Causes of UnfitnessCurativePalliative
  1. CGA indicates comprehensive geriatric assessment.

  2. More than 1 cause may be present in a single patient.

Comorbidity ≥3  
Geriatric syndrome  
 Neglect and abuse10
 Failure to thrive10
Activity in daily living  
Age ≥80 y78

Lymphoma-related pretreatment prognostic variables did not differ except for stage B, which was more frequent among unfit patients (Table 2). Fit patients obtained a significantly better outcome. The RR was 92.8% (39 of 42) compared with 52.3% (22 of 42) in unfit patients (Fisher exact test: P < .0001), the 2-year actuarial PFS was 73.4% compared with 21.7% (log-rank analysis: P < .0001), and OS was 77.6% compared with 23.8% (log-rank analysis: P < .0001).

Table 2. Characteristics and Outcome of Patients Classified as Fit or as Unfit According to CGA
CGA CategoryFitUnfitP
  • CGA indicates comprehensive geriatric assessment; SEM, standard error of the mean; NS, nonsignificant; IPI, International Prognostic Index; CR, complete response; PR, partial response.

  • *


No. of patients42 (50%)42 (50%) 
Mean age (±SEM)70.8 ± 0.8676.1 ± 0.83<.0001
Ann Arbor stage III-IV28 (66%)27 (64%)NS
Stage B9 (21%)18 (43%).03
IPI risk class*24 (57%)29 (69%)NS
Response rate39/42 (93%)22/42 (52%)<.0001
CR26 (62%)16 (38%) 
PR13 (31%)6 (14%) 
Overall survival (median)Not reached8 mo<.0001

On the basis of clinical judgment, all patients classified as fit were actually treated aggressively. An extrahematological toxicity World Health Organization grade ≥3 was recorded in 3 (7.1%) patients (pneumonitis + neuropathy + cardiac failure grade 3 in 1; gastrointestinal hemorrhage grade 3 in 1, and fatal septic shock in 1). In addition, 20 of 42 patients classified as unfit by CGA were also considered able to tolerate aggressive treatment with curative intent. They actually received a dose intensity of anthracycline, cyclophosphamide, and vincristine/vinblastine >80% of the planned dose intensity. The presenting characteristics of this subgroup of unfit patients treated aggressively were similar to those of unfit patients receiving palliation, regarding demographic features, lymphoma characteristics, and causes of “unfitness” (Table 3). As shown in Figure 2, their outcome was as poor as that of unfit patients receiving palliation only. The 2-year OS was 19.8% among unfit patients treated intensively and 26.1% among unfit patients receiving palliation (log-rank analysis: P = .85). Lymphoma rather than toxicity was the main cause of failure/death in both subgroups. It was present at death in 8 of 13 unfit patients treated intensively and in 11 of 15 receiving palliation. Toxicity was the cause of death in 4 patients treated intensively (respiratory failure, pneumonitis [2], sepsis, and liver failure) and in 4 patients receiving palliation (cardiac failure, multiorgan failure, pneumonitis, stroke). Within the subgroup of unfit patients treated with curative intent, there was no single clinical characteristic that seemed to be associated with a better outcome. In particular, 7 patients classified as unfit only because of age >80 years and receiving aggressive treatment had a poor outcome, similar to that of other unfit patients, 5 of them dying of lymphoma.

Figure 2.

Actuarial overall survival curves of elderly patients with diffuse large cell lymphoma classified as fit or unfit according to comprehensive geriatric assessment are shown. Unfit patients are subdivided into 2 subgroups according to the type of treatment received (log-rank analysis: P < .0001). pall indicates palliation; intens, intensive treatment.

Table 3. Characteristics and Outcome of Patients Classified as Unfit by CGA Subdivided According to the Type of Treatment Actually Received Based on Clinical Judgment
  • CGA indicates comprehensive geriatric assessment; SEM, standard error of the mean; NS, nonsignificant; IPI, International Prognostic Index; ADL, activity of daily living; CR, complete response; PR, partial response.

  • *


No. of unfit patients2022 
Mean age (±SEM)75.2 ± 1.2177.3 ± 1.26NS
Ann Arbor stage III-IV1512NS
Stage B108NS
IPI risk class*1514NS
Cause of unfitness   
Age >80 y78NS
Geriatric syndrome21NS
Therapy administered   
No therapy03 
Surgery (stage I)02 
Rituximab alone01 
Chemotherapy without anthracycline07 
Chemotherapy with anthracycline200 
Response rate11/20 (55%)11/22 (50%)NS
CR6 (30%)9 (41%) 
PR5 (25%)2 (9%) 
Overall survival (median)8 mo7 moNS

When comparing the OS of the 62 patients identified as able to receive intensive treatment by clinical decision with that of the 42 patients identified as fit by CGA, fit patients fared significantly better, considering both patients treated after 2003 (log-rank analysis: P = .049) and historical patients receiving intensive treatment between 1995 and 2002 at our institution (log-rank analysis: P = .027) (Fig. 3).

Figure 3.

Actuarial overall survival curves of elderly patients with diffuse large cell lymphoma receiving intensive treatment with curative intent are shown. Patients who were fit according to comprehensive geriatric assessment (CGA) had a significantly better survival compared with patients treated intensively based on clinical judgment both before 2003 (log-rank analysis: P = .027) and from 2003 to 2006 (log-rank analysis: P = .049).


Combination chemotherapy with standard-dose CHOP with rituximab is currently the treatment of choice in elderly patients with DLCL.11 Reduced-dose or modified regimens adapted to elderly patients have obtained inferior results. Therefore, the identification of those elderly patients whose global health status is fair enough to make it likely that they could benefit from chemotherapy treatment with curative intent has become of paramount importance. Age and performance status, according to Eastern Cooperative Oncology Group12 or Karnofsky score,13 are currently the most frequently used criteria to select patients for standard chemotherapy at full doses, but the final decision is generally left to the clinical judgment of the attending physician. Indeed, these parameters are largely inaccurate in describing the characteristics of elderly patients, and there is a large body of geriatric literature that tries to address this problem by designing specific CGA scores.

In the present study, a simplified method of CGA was applied to patients with DLCL to analyze if it could be used as an objective tool for the prospective identification of patients able to tolerate full-dose chemotherapy. The assessment test was easy to perform and did not require more than 15 minutes.

The geriatric parameters considered were evenly distributed among the subgroups of patients receiving aggressive or palliative treatment. Within the limitations of the small numbers of patients in some subgroups, no CGA parameters appeared to correlate with a different outcome. Results showed that CGA was more conservative than the clinical judgment applied over the course of a decade at our institution in assigning patients to the subgroup considered fit for full-dose treatment. It selected only 50% of the patients, whereas by clinical judgment approximately 75% of consecutively diagnosed patients were considered fit enough to benefit from an aggressive treatment approach with curative intent. Because this proportion has not changed for the past 10 years, it did not appear to depend either on the type of curative treatment or supportive measures available, or on the concurrent performance of CGA, which was indeed never used as a criterion for treatment decision.

Patients classified as fit by CGA had a very satisfactory clinical outcome using CHOP or CHOP-like regimens plus rituximab. The 2-year PFS and OS rates of 73.4% and 77.6%, respectively, did not substantially differ from the results achieved in younger patients with the same treatment.14-16 Moreover, major toxic effects were limited, and only 1 patient died of treatment-related toxicity.

Conversely, patients considered unfit by CGA had a poorer outcome irrespectively of the type of treatment actually received. Unexpectedly, the subgroup of unfit patients who were clinically considered able to benefit from full-dose treatment with curative intent had the same outcome as patients given palliation. The actual dose intensity received by unfit patients treated aggressively was >80% and could therefore not account for their poor outcome. Notably, toxicity but also lymphoma itself significantly contributed in causing the poor outcome of unfit patients receiving full-dose treatment, and in proportions similar to those of the subgroup receiving palliation. This suggests that the curability of DLCL may not only depend on its biological characteristics, but also on as yet unrecognized patient-related factors, which may be linked particularly to the parameters analyzed by CGA. Indeed, within unfit patients, no differences in pretreatment prognostic characteristics other than CGA parameters were found that could distinguish between the 2 subgroups.

In conclusion, the CGA used in our study appears to be a more effective and objective tool than clinical decision making to identify those patients who can safely be treated with full-dose, aggressive immunochemotherapy and can achieve an outcome similar to that of younger DLCL patients. However, these data should be validated in larger prospective randomized studies.

The survival of the subgroup of fit patients identified by CGA was significantly better than that of patients treated with curative intent based on clinical judgment, as applied at our institution. Conversely, patients identified by CGA as unfit seem to represent a subgroup in which the use of full-dose anthracycline-based treatment cannot improve the outcome compared with palliative treatment, both because of poor treatment tolerance and because of lymphoma refractoriness.

Conflict of Interest Disclosures

The authors made no disclosures.