Anemia of cancer in intermediate-grade non-Hodgkin's lymphoma.

ABSTRACT

Background. Current literature suggests that anemia at baseline is an important adverse prognostic factor for lymphoma-related outcomes. We evaluated the prevalence, risk factors, and prognostic value of anemia in patients with intermediate-grade non-Hodgkin's lymphoma (IGNHL) treated in a community-based practice.

Methods. The retrospective sample included 591 patients who had IGNHL diagnosed between 1993 and 1999 and who were subsequently treated with CHOP chemotherapy. Anemia was defined as a hemoglobin (Hb) value <12 g/dL.

Results. Anemia was present in 193 of 546 patients (35.3%). Baseline anemia was significantly associated with age >60, extranodal sites [grater than or equal to]2, Ann Arbor stage III or IV, elevated lactate dehydrogenase (LDH) level, B symptoms, and histology type. Baseline Hb was also a significant predictor of nonresponse to chemotherapy.

Conclusions. Our study results support previous findings of a high prevalence of anemia in cancer patients before cytotoxic therapy and the adverse impact that baseline anemia has on response to chemotherapy.

**********

THE PREVALENCE of anemia in IGNHL is not as well documented in community-based practice as it is in randomized clinical trials. Researchers have identified mediators of the inflammatory response such as interleukin 1, [gamma] interferon, and tumor necrosis factor that inhibit erythropoiesis. (1,2) Other factors implicated in this regard include bone marrow involvement, abnormal iron reutilization, inappropriately low serum erythropoietin levels for the degree of anemia, and a decrease of bone marrow response to erythropoietin. (3-7)

The prognostic value of bone marrow involvement in outcomes associated with NHL are established in the literature. (8) However, relatively few studies have evaluated anemia at diagnosis as a prognostic indicator in patients with NHL. A recent study that retrospectively evaluated anemic patients with NHL concluded that anemia was an important adverse prognostic factor in lymphoma patients. (2) In addition, risk factors associated with anemia at baseline in patients with NHL were identified as age, stage at diagnosis, extranodal site, serum albumin level, bone marrow involvement, and histology. Other investigators have also found that anemic cancer patients have a higher relapse rate or even a higher mortality rate than nonanemic patients (with similar grade or stage of disease). (9-14) However; some of these studies were conducted in anemic patients treated with radiotherapy, not chemotherapy. (9,10,13,14) In addition, these studies also included different tumor types such as ovarian, lung, and the glottic lary nx. (9-14)

The purpose of our study was to evaluate the prevalence and severity of anemia in patients with IGNHL in the community-based practice, before treatment with CHOP (cyclophosphamide, hydroxydaunomycin [doxorubicin], Oncovin [vincristine], and prednisone). We also explored clinical characteristics common among baseline anemic patients, and we evaluated the impact of pretreatment anemia on response to chemotherapy.

MATERIALS AND METHODS

Study Design and Patient Selection

We conducted a retrospective evaluation of patient medical records from 1991 through 1999 at 12 diverse oncology practice sites across the United States, including 2 academic centers, 5 integrated hospital systems, and 5 community practices. To be eligible for inclusion in the study, a patient should have been treated for IGNHL (stages I to IV), be at least 18 years of age, and have received only one chemotherapy regimen within 3 years of the studied chemotherapy course. To identify patterns of care in community-based practice, patients on a clinical treatment protocol were excluded from this study. Patients with other primary invasive cancers, receiving another course (ie, relapses and retreatment) of chemotherapy within 3 years of Oncology Practice Pattern Study initiation, or testing positive for human immunodeficiency virus infection, were excluded from the study.

A standard case report form was used to extract data from the patient medical records. Data extraction was conducted under the supervision of the physician principal investigator at each site. Collected data elements included the following: (1) patient characteristics including age, sex, race, comorbidity, size (height, weight, and body surface area (BSA) at each cycle), cancer stage, histology type, extranodal site involvement, elevated LDH, and B symptoms; (2) complete blood cell counts with differential (when available); (3) previous and/or concurrent radiation therapy including courses, total doses, and dates delivered; (4) planned and delivered chemotherapy regimen (drugs, dose, route of administration, number and length of cycles); (5) history of any surgical treatment, (6) short-term treatment complications (eg, mucositis, febrile neutropenia, hospitalizations with event dates and relevant documentation, and laboratory results); and (7) response to chemotherapy (documented by the treating physician). Anemia treatment data were not available. The study data were extracted from the medical records of patients with IGNHL. This study focused on those 591 patients who received initial CHOP chemotherapy.

Diagnosis of Anemia

Baseline Hb level was defined as the earliest recorded Hb level after diagnosis of cancer. This usually occurred just before the initiation of the first cycle of chemotherapy. Anemia was defined by the National Cancer Institutes' grading scale. (15) This scale categorizes anemia into five grades based on Hb levels: grade 0 (within normal limits, [greater than or equal to]12.0 g/dL); grade 1 (mild, 11.9 to 10.0 g/dL); grade 2 (moderate, 9.9 to 8.0 g/dL); grade 3 (serious/severe, 7.9 to 6.5 g/dL); and grade 4 (life-threatening, <6.5 g/dL). Consequently, for the purpose of our analysis, anemia was defined as a Hb value <12 g/dL.

Measure of Response

Response to chemotherapy was collected from the clinical charts as complete response (CR), partial response (PR), and no response/progression (NR). For logistic regression modeling purposes, response was dichotomized into CR or PR versus NR.

Missing Values 

A measurement of Hb before CHOP therapy (baseline Hb) was not available for 45 patients. In addition, response to chemotherapy was not reported for 45 patients. For 2 patients, both measures were missing. In the initial analysis, patients with missing values were dropped from the two corresponding analyses. In the subsequent analysis, to evaluate the effect of missing outcome data on the results, patients without baseline Hb measures were conservatively assumed to have normal Hb levels and patients with missing response to chemotherapy were presented as a separate response category. Patients with missing data for independent variables were dropped from the analyses involving those variables. Independent variables with missing values included stage (49 patients), LDH (195 patients), and histology type (81 patients). In combination, 252 patients were missing at least one independent variable value.

Study Variables and Operational Definitions

Baseline patient characteristics included age, sex, BSA, treatment year, Ann Arbor stage, extranodal involvement, serum LDH level, comorbid conditions, presence of heart disease, presence of renal disease, previous radiation therapy, presence of B symptoms, and lymphoma histology type.

Age (<60 or [greater than or equal to]60 years) and sex were set as dichotomous variables. Body surface area was dichotomized as less than or greater than the sample median (1.88 [m.sup.2]). Treatment year was classified as either 1991 to 1994 or 1995 to 1999. Non-hodgkin's lymphoma was categorized as limited stage (Ann Arbor stages I or II)or advanced stage (Ann Arbor stages III or IV). Extranodal involvement was dichotomized as 0 to 1 or [greater than or equal to]2. The LDH level was categorized as either normal or elevated. Comorbid conditions were classified according to the Charlson Comorbidity Index (CCI), which incorporates ICD-9 coding. The number of comorbid conditions in the study sample ranged from 0 to 12. The CCI data were collapsed into two ordinal categories (0 and [greater than or equal to]1) to distinguish the patients with and without comorbid conditions. In addition to the overall evaluation of the comorbid conditions, heart disease (ICD9-CM codes 410, 411, 412, 414, 427, and 428) and renal disease (ICD9-CM codes 403, 404, 580 to 586, and 588) were analyzed separately and classified as either present or absent. It is commonly understood that both of these conditions (renal and heart disease) have an impact on anemia or are affected by anemia. Renal disease may result in decreased production of erythropoietin followed by a decrease in red blood cell count. Anemia, which results in poor blood oxygenation, may aggravate cardiac conditions such as congestive heart failure and angina. Previous radiation therapy was dichotomized regarding whether it was administered before chemotherapy. The variable B symptoms (which include recurrent fever, night sweats, or the loss of more than 10% of body weight) was classified as present or absent The International Working Formulation served as the basis for the lymphoma histology. (16) Finally, even though bone marrow involvement and performance status data were collected, they were considered incomplete and therefore were excluded from analysis.

Statistical Methods

Statistical analyses were done to assess the associations of the two outcomes, viz, anemia at baseline and response to chemotherapy, to various baseline patient characteristics. In the univariate analyses, continuity-adjusted chisquare tests were used to identify statistically significant associations between the two outcomes and various patient characteristics. For each outcome measure, the significant variables identified in the univariate analysis were entered into a stepwise multiple logistic regression. Each of these two analyses enabled us to identify a group of clinical risk factors that were associated with the two outcomes (baseline anemia/no anemia and chemotherapy response/no response). Two-way interactions were tested for significance. C-statistic (0.5 [less than or equal to] c [less than or equal to] 1) was used to measure the discriminating power of the models, and the Hosmer-Lemeshow lack-of-fit test was used to assess the adequacy of the model over the range of risk levels. All reported P valu es in the results section are two-sided. Analyses were done using SAS PC version 8.0.

RESULTS

Prevalence of Anemia

A total of 546 patients with available baseline Hb values in the study sites database were treated with CHOP from 1991 to 1999. About 35% (193 of the 546 patients) were anemic (Hb <12 g/dL) at baseline. Of these 193 anemic patients, 131 (67.9%) had Hb values between 10.0 and 11.99 (NCI grade 1; mild), 53 (27.5%) had Hb values between 8.0 and 9.99 (NCI grade 2; moderate), and 9 (4.7%) had Hb values <8 g/dL (NCI grades 3 and 4; severe to life-threatening). Data on patients with NCI grade 3 and 4 anemia were combined for the analysis, since there were only a small number of patients in these categories.

Patient Demographics

The median age of the study sample (546 patients) was 65.1 years, with 61.4% of the patients over 60 years old. About half of the patients were female and had advanced disease, 21.3% had [greater than or equal to]2 extranodal sites, and 45.6% had an elevated LDH. Fifty-seven percent of the patients had diffuse large histologic cell type. About 13% of the patients had heart disease, and only 5% had renal disease. Twenty percent of the population had B symptoms.

Patient Characteristic Associated With Baseline Anemia

In univariate analyses (Table 1), anemia was significantly associated with age over 60 (38.8% vs 29.9%; P = .0416), female sex (42.4% vs 28.7%; P < .0001), a lower than average BSA (43.4% vs 27.4%; P <.0001), stage III and IV NHL at diagnosis (41.8% vs 28.7%; P = .0031), number of extranodal sites [greater than or equal to]2 (43.5% vs 31.5%; P = .042), elevated LDH (51.5% vs 23.3%; P = .001), presence of renal disease (63.3% vs 33.7%; P = .002), presence of B symptoms (51.4% vs 31.3%; P < .0001), and certain histology types, specifically large cell diffuse and immunoblastic (38.5% and 56.0%, respectively; P < .0001). The only factors that tested nonsignificant were treatment year, CCI, the presence of heart disease, and previous radiation therapy.

Multiple Logistic Regression for Baseline Anemia

All significant factors identified in the univariate analyses were entered into a stepwise multiple logistic regression model (Table 2). Anemia was found to be significantly associated with female sex, elevated LDH, presence of B symptoms, advanced stage NHL (stage III or IV) at diagnosis, and large cell diffuse and immunoblastic histology types. There were no significant interactions between any of the variables included in the model. Controlling for the other factors in the model, women were 2.41 times more likely to present with baseline anemia than men (P = .0005), patients with elevated LDH were 2.62 times more likely to have baseline anemia than those with normal LDH (P = .0001), patients with B symptoms were 2.08 times more likely to have anemia than B asymptomatic patients (P = .0114), patients with advanced stage NHL were 1.87 times more likely to have anemia than those with early stage NHL (P = .0129), and finally, patients with large cell diffuse and immunoblastic histology types were 2.12 (P = .0066) and 4.10 (P = .0007) times more likely to have baseline anemia than those with lymphoma of other histology types. The value of c-statistic was 0.749.

To assess potential bias in excluding patients with missing baseline Hb measurement, a multiple logistic regression model was developed for a sample that includes those patients as nonevents (normal Hb). The results obtained were similar to those presented in Table 2.

Anemia and Response to Chemotherapy

Table 3 shows the relationship between baseline Hb values and response to chemotherapy as noted by the treating physician. Overall, anemic patients (Hb <12 g/dL) had a higher than expected percentage of no response/progression than nonanemic patients. Anemia at baseline was present for 46.1% to 51.7% of patients in the no response/progression group, as compared with 35% baseline anemia overall. The upper bound of 51.7% was obtained from Table 3 (A) by excluding patients with missing baseline Hb value or response to chemotherapy. The lower bound of 46.1% was obtained from Table 3 (B) assuming that patients with missing baseline Hb had normal Hb levels and including patients with missing response information as a separate category.

Multiple Logistic Regression for Response to Chemotherapy

A multiple logistic regression model (Table 4) was also used to identify significant factors associated with no response to chemotherapy (predicting NR versus CR/PR). The following patient characteristics were included as the predictor variables in the model: baseline anemia measured as Hb <12 g/dL, age, sex, BSA, cancer stage, number of extranodal sites, presence of heart and kidney disease, planned dose intensity, LDH, and previous radiation therapy. The results presented in Table 4 imply that the presence of baseline anemia was a significant risk factor for NR to CHOP chemotherapy (OR = 2.29; 95% CI 1.10-4.74) controlled for elevated LDH, advanced stage III and IV, and age 60. This led to the conclusion that baseline anemia is a prognostic factor for response to chemotherapy even after controlling for a subset of the five commonly used factors in the International Prognostic Index. Performance status and number of extranodal sites were only partially available and were excluded from this analysis.

DISCUSSION

The purpose of this retrospective study was to determine at baseline (time of cancer diagnosis before chemotherapy) the prevalence, risk factors, and prognostic value of anemia in a group of patients with IGNHL treated in community-based practice. Thirty-five percent of those in our study sample were anemic at baseline. This was consistent with findings in other studies. The literature reports the frequency of baseline anemia in low-grade lymphomas in the 29% to 37% range. (17-20) For intermediate-grade and high-grade lymphomas, the frequency ranges from 28% to 47%. (8-21-25) In our study, more than half of the patients had mild anemia at baseline. Anemia differentiated by the grade of NHL was not reported in the literature. As mentioned earlier, Moullet et a1 (8) evaluated the characteristics common among patients who were anemic at diagnosis. Our study results were similar to those of Moullet et al. (8) They found that a significantly larger percentage of anemic patients had the following characteristics: [greater than or equal to] 60 years of age (39%), stage III or IV disease (37%), elevated LDH (47%),presence of B symptoms (61%), and [greater than or equal to] 2 extranodal sites (47%). However, Moullet et al (8) did not evaluate other factors such as presence of kidney disease, previous radiation therapy, low BSA, and sex. Although our findings were consistent with those of Moullet et al, (8) our study focused only on patients with intermediate-grade lymphoma, whereas their study included patients with all three grades of NHL.

A major limitation of our study was that a significant number of patients were missing values for various variables used in the analyses. We conducted a sensitivity analysis to assess the impact of missing outcomes (baseline anemia and response to chemotherapy) on the reported results. Baseline anemia was present in 46.1% to 51.7% of the patients who had no response/progression. The reported models showed little sensitivity to missing baseline anemia and response. It seemed less plausible to assume or impute the values of the independent variables lymphoma stage, LDH, and histology, and therefore we chose to exclude these patients from the corresponding analyses.

There appears to be a lack of consistency in published literature documenting the relationship between anemia and histology subtypes. This may be driven by the fact that various studies had a different distribution of patients for different NHL grades. Moullet et al (8) showed that the incidence of anemia was low in indolent lymphomas (17% in follicular and 19% in marginal zone lymphomas) except for small lymphocytic or lymphoplasmacytoid (37%). Conlan et al (26) found that patients with high-grade lymphoma had a lower median Hb value and a higher prevalence of anemia than patients with intermediate-grade and low-grade lymphomas. In contrast, Bloomfield et al (18) found no significant difference among histology types. In our study, prevalence of anemia varied with histology subtypes. This was consistent with the published literature. Increased cytokine production may be responsible for the variation in anemia patterns based on histology subgroups. (8) In our study of intermediate-grade NHL, a larger percentage of anemic patients had the following histology subtypes: large cell immunoblastic (56%) and large cleaved or noncleaved cell/diffuse (38.9%). In addition, the smallest percentage of anemic patients had small cleaved/cell-diffuse histology.

There are conflicting reports in the literature regarding bone marrow involvement in anemia. Both Moullet et al (8) and Bloomfield et al (18) indicated that bone marrow involvement was significantly associated with anemia. Other investigators have reported that even though bone marrow involvement was associated with anemia, it was not at a significant level. (26,27) In our study, data regarding bone marrow involvement were incomplete and thus were excluded from analysis. This could have had an impact on the results.

Multiple studies have shown that anemia may affect response to radiotherapy. Although the effect of oxygen delivery to the tumor is unknown, evidence shows that it enhances the efficacy of radiotherapy. (28) Various studies have determined that anemia before chemotherapy or radiotherapy has a negative correlation with survival (20,22,23,26,29,34) and also decreases response to therapy. (17,31) Limited data in the literature document the impact of pretreatment anemia on response to chemotherapy Morel et al (17) evaluated the factors that influence complete remission in patients with NHL and acute lymphoblastic leukemia and found that age >40, high LDH level, low Hb level (<10 g/dL), and two or more extranodal sites were associated with poor response. However. complete remission was measured over a longer period after chemotherapy. Additionally, these patients were treated with four different chemotherapy regimens. In contrast, our study included only patients with IGNHL treated with CHOP, and response to chemo therapy was documented from the chart and follow-up. Response to chemotherapy was inversely associated with baseline Hb. Anemia was significantly associated with the factors listed. There is a potential for bias in our study because chemotherapy response was a subjective evaluation by the treating physician. Nevertheless, the results do show that there is consistency with current literature and that significant association exists between baseline anemia and response to chemotherapy. The mechanism of this association has yet to be determined and wan-ants further evaluation. Our study results underscore the importance of identifying risk factors among patients who are anemic before receiving cancer therapy In addition. we found that baseline anemia could be considered as a prognostic factor for response to chemotherapy Consequently, it seems that baseline anemia in IGNHL needs to be identified and appropriately managed.

TABLE 1

Patient Characteristics Associated With Baseline Anemia (Hemoglobin <12
g/dL): Results from Univariate Analyses

                                                    Patients With
              Patient             Total No. (%) *    Anemia (%) +
          Characteristic             (N = 546)     (N = 193 [35.3])

Age
    <60                             211 (30.6)         63 (29.9)
    [greater than or equal to]60    335 (61.4)        130 (38.8)
Sex
    Male                            282 (51.6)         81 (28.7)
    Female                          264 (48.4)        112 (42.4)
Body surface area
    <Median (1.88 [m.sup.2])        272 (49.8)        118 (43.4)
    [greater than or equal to]      274 (50.2)         75 (27.4)
     Median
Treatment year
    1991-1994                       188 (34.4)         68 (36.2)
    1995-1999                       358 (65.6)        125 (34.9)
Stage at diagnosis
    I-II                            251 (49.8)         72 (28.7)
    III-IV                          249 (50.2)        104 (41.8)
Number of extranodal sites
    <2                              340 (78.7)        107 (31.5)
    [greater than or equal to]2      92 (21.3)         40 (43.5)
Lactate dehydrogenase
    Normal                          202 (54.4)         47 (23.3)
    Elevated                        169 (45.6)         87 (51.5)
Charlson Comorbidity Index
    0                               350 (64.1)        115 (32.9)
    [greater than or equal to]1     196 (35.9)         78 (39.8)
Heart disease
    Present                          73 (13.4)         32 (43.8)
    Absent                          473 (86.6)        161 (34.0)
Renal disease
    Present                          30 (5.5)          19 (63.3)
    Absent                          516 (94.5)        174 (33.7)
Previous radiation therapy
    Yes                              14 (2.6)           7 (50.0)
    No                              532 (79.7)        186 (35.0)
B symptoms
    Yes                             111 (20.3)         57 (51.4)
    No                              435 (79.7)        136 (31.3)
Lymphoma histology
    Follicular large cell            38 (8.0)           9 (23.7)
    Small cleaved cell diffuse       24 (5.1)           3 (12.5)
    Mixed small and large            72 (15.2)         20 (27.8)
    cell diffuse
    Large cleaved or noncleaved     288 (60.9)        112 (38.9)
    cell/diffuse
    Large cell immunoblastic         50 (10.6)         28 (56)
    Other                             1 (0.2)           0 (0)


              Patient               P
          Characteristic          Value **

Age
    <60                            .0416
    [greater than or equal to]60
Sex
    Male                          <.0001
    Female
Body surface area
    <Median (1.88 [m.sup.2])      <.0001
    [greater than or equal to]
     Median
Treatment year
    1991-1994                      .844
    1995-1999
Stage at diagnosis
    I-II                           .0031
    III-IV
Number of extranodal sites
    <2                             .0420
    [greater than or equal to]2
Lactate dehydrogenase
    Normal                        <.0001
    Elevated
Charlson Comorbidity Index
    0                              .125
    [greater than or equal to]1
Heart disease
    Present                        .103
    Absent
Renal disease
    Present                        .002
    Absent
Previous radiation therapy
    Yes                            .245
    No
B symptoms
    Yes                           <.0001
    No
Lymphoma histology                <.0001
    Follicular large cell
    Small cleaved cell diffuse
    Mixed small and large
    cell diffuse
    Large cleaved or noncleaved
    cell/diffuse
    Large cell immunoblastic
    Other

* Column percent.

+ Row percent.

** Continuity adjusted chi-square test P values (except for histology,
which was not adjusted).
TABLE 2

Multiple Logistic Regression Results for the Risk of Anemia at Baseline
(N = 349)

                                         Anemia Event

Risk Factors                     OR        95 % CI   P Value

Female sex                      2.41      1.47-3.96   .0005
Elevated lactate dehydrogenase  2.62      1.60-4.28   .0001
B symptoms present              2.08      1.18-3.68   .0114
Stage III-IV                    1.87      1.14-3.06   .0129
Lymphoma histology
 Large cell diffuse             2.12      1.23-3.65   .0066
 Large cell immunoblastic       4.10      1.82-9.23   .0007
Overall fit
 Log-likelihood test                                  .0001
 C statistic                                          .749
 Hosmer-Lemeshow                                      .3030

Only significant variables are listed.

OR = Odds ratio, CI = Confidence interval.
TABLE 3

Baseline Hemoglobin Values and Response to Chemotherapy

(A) Excluding Missing Hemoglobin and Response Data

           Baseline              Complete     Partial    No Response/
       Hemoglobin Level          Response     Response   Progression
            (g/dL)               No. (%)      No. (%)      No. (%)

[greater than or equal to]12.0  215 (65.2)    87 (26.4)   28 (8.5)
          10.0-11.9              68 (57.63)   30 (25.4)   20 (17.0)
           8.0-9.9               19 (40.4)    20 (42.6)    8 (17.0)
             <8.0                 3 (37.5)     3 (37.5)    2 (25.0)
               Total            305 (60.7)   140 (27.8)   58 (11.5)

           Baseline             Missing
       Hemoglobin Level         Response      Total
            (g/dL)              No. (%)      No. (%)

[greater than or equal to]12.0             330 (65.6)
          10.0-11.9                        118 (23.5)
           8.0-9.9                          47 (9.3)
             <8.0                            8 (1.6)
               Total                       503 (100.0)

(B) Including Missing Hemoglobin
and Response Data

           Baseline              Complete     Partial    No Response/
       Hemoglobin Level          Response     Response   Progression
            (g/dL)               No. (%)      No. (%)      No. (%)

      Missing hemoglobin         25 (55.6)    11 (24.4)    7 (15.6)
[greater than or equal to]12.0  215 (60.9)    87 (24.7)   28 (7.9)
          10.0-11.9              68 (51.9)    30 (22.9)   20 (15.3)
           8.0-9.9               19 (35.9)    20 (37.7)    8 (15.1)
             <8.0                 3 (33.3)     3 (33.3)    2 (22.2)
               Total            330 (55.8)   151 (25.6)   65 (11.0)

           Baseline             Missing
       Hemoglobin Level         Response      Total
            (g/dL)              No. (%)      No. (%)

      Missing hemoglobin         2 (4.4)    45 (7.6)
[greater than or equal to]12.0  23 (6.5)   353 (59.7)
          10.0-11.9             13 (9.9)   131 (22.2)
           8.0-9.9               6 (11.3)   53 (9.0)
             <8.0                1 (11.1)    9 (1.5)
               Total            45 (7.6)   591 (100.0)

A: Chi-square P value = .007.

B: Chi-square P value = .046.
TABLE 4

Predicitive Factors Associated With No Response to Chemotherapy
(N = 332)

                                               No Response

Risk Factors                       Odds Ratio      95% CI    P Value

Baseline anemia
    Baseline hemoglobin <12 g/dL      2.29       1.10-4.74    .0263
LDH by Stage interaction
    Elevated LDH, Advanced stage      7.97       1.73-36.82   .0079
    Elevated LDH, Limited stage      11.29       2.43-52.41   .0020
    Normal LDH, Advanced stage        5.88       1.26-28.23   .0268
Age [greater than or equal to]60      1.75       0.82-3.74
  years
Overall fit
    Log-likelihood test                                       .0097
    C statistic                                               .743
    Hosmer-Lemeshow                                           .7514

CI = Confidence interval

LDH = lactate dehydrogenase.

References

(1.) Means RT Jr, Dessypris EN, Krantz SB: Inhibition of human erythroid colony-forming units by intcrleukin-1 is mediated by y-interfcron. J Cell Physiol 1992; 150:59-64

(2.) Mamus SW, Beck-Schroeder S, Zanzani ED: Suppression of normal human erythropoiesis by y interferon in vitro. J Clin Invest 1985; 75:1496-1503

(3.) Spivak JL: Cancer-related anemia: its causes and characteristics. Semin Oncol 1994; 21:3-8

(4.) Miller CB, Jones RJ, Piantadosi S, et al: Decreased erythropoietin response in patients with the anemia of cancer. N Engl J Med 1990; 322:1689-1692

(5.) Zucker S: Anemia in cancer. Cancer Invest 1985; 3:249-260

(6.) Jelkmann W, Page H, Wolff M, et al: Monokines inhibiting erythropoietin production in human hepatoma cultures and in isolated perfused rat kidneys. Life Sci 1992; 50:301-308

(7.) Faquin WC, Schneider TJ, Goldberg MA, et al: Effect of inflammatory cytokines on hypoxia-induced erythropoietin production. Blood 1992; 79:1987-1994

(8.) Moullet I, Salles G, Ketterer N, et al: Frequency and significance of anemia in non-Hodgkin's lymphoma patients. Ann Oncol 1998; 9:1109-1115

(9.) Reed WR, Hussey DH, DeGowin RL: Implications of the anemia of cancer disorders in patients anticipating radiotherapy. Am J Med Sri 1994; 308:9-15

(10.) Overgaard J, Hansen HS, Jorgensen K, et al: Primary radiotherapy of larynx and pharynx carcinoma, an analysis of some factors influencing local control and survival. Int J Radial Oncol Biol Phys 1986; 12:515-521

(11.) Obermair A, Petru E, Windbichler G, et al: Significance of pretreatment serum hemoglobin and survival in epithelial ovarian cancer. Oncol Rep 2000; 7:639-644

(12.) Obermair A, Handisurya A, Kaider A, et al: Influence of pretreatment serum hemoglobin levels on survival in epithelial ovarian cancer: a prospective review. Cancer 1998; 83:726-731

(13.) Fein DA, Lee WR, Hanton AL, et al: Pretreatment hemoglobin level influences local control and survival of TI-T2 squamous cell carcinomas of the glottic larynx. J Clin Oncol 1995; 13:2077-2083

(14.) Kumar P: Impact of anemia in patients with head and neck cancer. Oncologist 2000; 5(suppl 2):13-18

(15.) Groopman LE, Itri LM: Chemotherapy-induced anemia in adults: incidence and treatment. J Natl Cancer Inst 1999; 91:1616-1634

(16.) The Non-Hodgkin's Lymphoma Pathologic Classification Project. National Cancer Institute sponsored study of classifications of non-Hodgkin's lymphomas. Cancer 1982; 49:2112-2135

(17.) Morel P, Lepage E, Brice P, et al: Prognosis and treatment of lymphoblastic lymphoma in adults: a report on 80 patients. J Clin Oncol 1992; 7:1078-1085

(18.) Bloomfield CD, McKenna RW, Brunning RD, et al: Significance of hematological parameters in the non-Hodgkin's malignant lymphomas. Br J Hematol 1976; 32:41-46

(19.) Cabanillas F, Smith T, Bodey GP, et al: Nodular malignant lymphomas. factors affecting complete response rate and survival. Cancer 1979; 44:1983-1989

(20.) Gallagher GJ, Gregory WM, Jones AE, et al: Follicular lymphoma: prognostic factors for response and survival. J Clin Oncol 1986; 4:1470-1480

(21.) Coiffier B, Gisselbrecht C, Herbrecht R, et al: LNH-84 regimen: a multicenter study of intensive chemotherapy in 737 patients with aggressive malignant lymphoma. J Clin Oncol 1989; 7:1018-1026

(22.) Cowan RA, Jones M, Harris M, et al: Prognostic factors in high and intermediate-grade non-Hodgkin's lymphoma. Br J Cancer 1989; 50:276-282

(23.) Dixon DO, Neilan B, Jones SE, et al: Effect on age on therapeutic outcome in advanced diffuse histiocytic lymphoma: the Southwest Oncology Group experience. J Clin Oncol 1986; 4:295-305

(24.) Kaplan WD, Jochelson MS, Herman TS, et al: Gallium-67 imaging: a predictor of residual tumor viability and clinical outcome in patients with diffuse large-cell lymphoma. J Clin Oncol 1990; 8:1966-1970

(25.) Kwak LW, Halpern J, Oshen RA, et al: Prognostic significance of actual dose intensity in diffuse large-cell lymphoma: results of a tree-structured survival analysis. J Clin Oncol 1990; 8:963-977

(26.) Conlan MG, Armitage JO, Bast M, et al: Clinical significance of hematologic parameters in non-Hodgkin's lymphoma at diagnosis. Canacer l99l; 67:1389-1395

(27.) Stein RS, Ultmann JE, Byrne GB, et al: Bone marrow involvement in non-Hodgkin's lymphoma. implications for staging and therapy. Cancer 1976; 37:629-636

(28.) Molls M, Stadler P, Becker A, et al: Relevance of oxygen in radiation therapy. mechanisms of action, correlation to low hemoglobin levels. Strahlenther Onkol 1998; 174:13-16

(29.) Romaguera KE, McLaughlin, Nort L, et al: Multivariate analysis of prognostic factors in stage IV follicular low-grade lymphoma: a risk model. J Clin Oncol 1991; 9:762-769

(30.) Leonard RCF, Hayward RL, Prescott RJ, et al: The identification of discrete prognostic groups in low-grade non-Hodgkin's lymphoma. Ann Oncol 1991; 2:655-662

(31.) Bremnes RM, Bremnes Y, Donnem T: High-grade non-Hodgkin's lymphoma treated in northern Norway. treatment, outcome, and prognostic factors. Acta Oncol 1999; 38:117-124

(32.) Ben-Bzra J, Burke JS, Schwartz WG, et al: Small lymphocytic lymphoma: a clinicopathologic analysis of 268 cases. Blood 1987; 70:1394-1399

(33.) Maksymiuk AW, Bratvold Y, Ezzat W, et al: Non-Hodgkin's lymphoma in Saskatchewan: a review of 10 years' experience. Cancer 1994; 73:711-779

(34.) Stewart WP, Crowther D, McWilliam LJ, et al: Maintenance chlorambucil after CVP in the management of advanced stage, low-grade histologic type non-Hodgkin's lymphoma: a randomized prospective study with an assessment of prognostic factors. Cancer 1988; 61:441-447

RELATED ARTICLE: KEY POINTS

* The prevalence of anemia in intermediate-grade non-Hodgkin's lymphoma (IGNHL) is not as well documented in community-based practice as it is in randomized clinical trials.

* Current literature suggests that anemia at baseline is an important adverse prognostic factor for lymphoma-related outcomes.

* Anemia was prevalent in the study population. Baseline anemia was significantly associated with age >60, extranodal sites [grater than or equal to]2, Ann Arbor stage III or IV, elevated lactate dehydrogenase level, B symptoms, and histology type. Baseline hemoglobin was also a significant predictor of nonresponse to chemotherapy.

* Patients at risk for baseline anemia in IGNHL may be identified and appropriately managed.

Insect Repellents

In addition to being a nuisance, insects and mosquitoes transmit diseases to more than 700 million people worldwide every year; and outbreaks of encephalitis and West Nile virus have caused illness and death in the US in the past 3 years. The best way to protect oneself from arthropod bites is to wear protective clothing, use insect repellent, and stay away from infested habitats. Mark S. Fradin, MD, and John F. Day, PhD, have compared the effectiveness of 16 insect repellent products available in the US. The full paper is published in the New England Journal of Medicine, Vol. 347, July 4, 2002, pages 13-18. We are pleased to reproduce a table summarizing their results and acknowledge permission granted from the New England Journal of Medicine.

TABLE

Protection Times of Insect Repellents *

                                     Active Ingredient
Product                              and Concentration

OFF! Deep Woods (Sc Johnson)         DEET, 23.8%
Sawyer Controlled Release (Sawyer)   DEET, 20%
OFF! Skintastic (SC Johnson)         DEET, 6.65%
Bite Blocker for Kids (HOMS)         Soybean oil, 2%
OFF! Skintastic for kids (SC         DEET, 4.75%
 Johnson)
Skin-So-Soft Bug Guard Plus (Avon)   IR3535, 7.5%
Natrapel (Tender)                    Citronella, 10%
Herbal Armor (microencapsulated)     Citronella, 12%
 (All Terrain)                       peppermint oil, 2.5%;
                                     cedar oil, 2%;
                                     lemongrass oil, 1%;
                                     geranium oil, 0.05%
Green Ban for people (Mulgum Hollow  Citronella, 10%;
 Farm)                               peppermint oil, 2%
Buzz Away (Quantum)                  Citronella, 5%
Skin-So-Soft Bug Guard (Avon)        Citronella, 0.1%
Skin-So-Soft Bath Oil (Avon)         Uncertain (n)
Skin-So-Soft Bath Moisturizing       Citronella, 0.05%
 Suncare (Avon)
Gone Original Wristband (Solar       DEET, 9.5%
 Gloooow)
Repello Wristband (Repello           DEET, 9.5%
 Products)
Gone Plus Repelling Wristband        Citronella, 25%
 (Solar Gloooow)

                                        Complete-Protection Time (min)
Product                               Mean                  Range

OFF! Deep Woods (Sc Johnson)         301.5[+ or -]37.6      200-360
Sawyer Controlled Release (Sawyer)   234.4[+ or -]31.8      180-325
OFF! Skintastic (SC Johnson)         112.4[+ or -]20.3      90-170
Bite Blocker for Kids (HOMS)         94.6[+ or -]42.0       16-195
OFF! Skintastic for kids (SC         88.4[+ or -]21.4       45-120
 Johnson)
Skin-So-Soft Bug Guard Plus (Avon)   22.9[+ or -]11.2       10-60
Natrapel (Tender)                    19.7[+ or -]10.6       7-60
Herbal Armor (microencapsulated)     18.9[+ or -]13.3       1-55
 (All Terrain)



Green Ban for people (Mulgum Hollow  14.0[+ or -]11.3       1-45
 Farm)
Buzz Away (Quantum)                  13.5[+ or -]7.5        5-30
Skin-So-Soft Bug Guard (Avon)        10.3-7.9               1-30
Skin-So-Soft Bath Oil (Avon)         9.6[+ or -]8.8         1-30
Skin-So-Soft Bath Moisturizing       2.8-3.4                1-15
 Suncare (Avon)
Gone Original Wristband (Solar       0.3[+ or -]0.2         0.17-1.33
 Gloooow)
Repello Wristband (Repello           0.2[+ or -]0.08        0.17-0.63
 Products)
Gone Plus Repelling Wristband        0.2[+ or -]0.09        0.17-0.48
 (Solar Gloooow)

                                     Category of
Product                              Protection +

OFF! Deep Woods (Sc Johnson)              A
Sawyer Controlled Release (Sawyer)        B
OFF! Skintastic (SC Johnson)              C
Bite Blocker for Kids (HOMS)              D
OFF! Skintastic for kids (SC              D
 Johnson)
Skin-So-Soft Bug Guard Plus (Avon)        E ++
Natrapel (Tender)                         E ++
Herbal Armor (microencapsulated)          E (ss)
 (All Terrain)



Green Ban for people (Mulgum Hollow       E
 Farm)
Buzz Away (Quantum)                       E
Skin-So-Soft Bug Guard (Avon)             E
Skin-So-Soft Bath Oil (Avon)              E
Skin-So-Soft Bath Moisturizing            F
 Suncare (Avon)
Gone Original Wristband (Solar            G
 Gloooow)
Repello Wristband (Repello                H
 Products)
Gone Plus Repelling Wristband             H
 (Solar Gloooow)

* Plus-minus values are the means [+ or -] SD of the time to the first
bite in the tests of all 15 subjects.

DEET denotes N,N-diethyl-3- methylbenzamide (formerly known as
N,N-diethyl-m-toluamide).

HOMS Home Operations and Management Systems, and [R3535 ethyl
butylacetylaminopropionate.

+ The mean complete-protection time of each repellent was significantly
different (P<0.05 by analysis of variance and Tukey's  test) from those
of all repellents  is in different categories of  protection (A, B, C,
D, E, F, G, and H).

++ The complete-protection time also different significantly from those
of Buzz Away, Skin-So- Soft Bug Guard, and Skin-So-Soft Bath Oil.

(ss)The complete-protection time also differed significantly from those
of Skin-So-Soft Bug Guard and Skin-So-Soft Bath Oil.

(n)This product contains mineral oil, isopropyl palmitate, dicapryl
adipate, fragrance, dioctyl sodium sulfosuccinate, butylated
hydroxytoluene, and carrot oil.

From One Health Plan, Atlanta, Ga; Professional Corporation, Staten Island, NY; Medical Affairs, Pharmacoeconomics, Amgen Inc, Thousand Oaks, Calif; and AMF Consulting Inc, Los Angeles, Calif.

Supported in part by Amgen Inc, Thousand Oaks, Calif.

Reprint requests to Thomas J. Morrow, MD, 320 Pilgrimage Point, Alpharetta, GA 30022.