Neutropenic complications in advanced-stage non-Hodgkin's lymphoma : Implications for the use of prophylactic recombinant human granulocyte-colony stimulating factor (G-CSF)

Objectives: To determine the incidence of neutropenic complications in patients receiving chemotherapy for advanced-stage non-Hodgkin's lymphoma (NHL), to predict which patients are at high risk for neutropenic complications and to develop an economic model for subsequent testing to assess the potential cost-effectiveness of prophylactic treatment with recombinant human granulocyte-colony stimulating factor (G-CSF).

Design: Retrospective chart review.

Patients: Forty-two patients with advanced-stage NHL treated at the Tom Baker Cancer Centre, Calgary, between Jan. 1, 1992, and Dec. 31, 1993.

Outcome measures: Neutropenic complications including incidence of febrile neutropenic events, documented infections, and chemotherapy dose delays or dose reductions.

Results: Of the 42 patients, 8 (19%) experienced febrile neutropenic events and 18 (43%) required chemotherapy dose modifications (delays or reductions or both) because of neutropenia. Fifteen patients (36%) were identified as being at high risk for neutropenic complications and may have benefited from the administration of prophylactic G-CSF. An economic model developed to assess the potential cost-effectiveness of prophylactic G-CSF therapy estimated that, for high-risk patients, the theoretical incremental cost per life year saved was $3300.

Conclusions: Febrile neutropenia and infection cause significant morbidity and mortality in patients receiving combination chemotherapy for the treatment of advanced-stage NHL. Secondary prophylactic G-CSF therapy has been proven to decrease the incidence of febrile neutropenia and infection in these patients. Considering the reduction in neutropenic complications and resulting increase in chemotherapy dose intensity received by patients on G-CSF, the theoretical incremental cost per life year saved of $3300 with G-CSF therapy is relatively low compared with other medical interventions.

Infection is the major complication for patients receiving combination chemotherapy for advanced-stage non-Hodgkin's lymphoma (NHL). In patients with severe neutropenia (absolute neutrophil count [ANC] less than 500/?L), the morbidity associated with infection is substantial, with mortality approaching 10%.1-4 Fever in a patient with neutropenia is often considered a medical emergency and remains an indication for hospitalization and antibiotic therapy.5,6 An estimated 75% of patients who complete chemotherapy for NHL undergo treatment delays of at least 7 days because of neutropenia.1

Recombinant human granulocyte-colony stimulating factor (G-CSF) has been proven to decrease the severity and duration of neutropenia and therefore the incidence of neutropenic complications in patients receiving chemotherapy.1,7-9 G-CSF is generally well tolerated, the main side effect being mild bone pain; there is no evidence that G-CSF may stimulate the growth of malignant cells in vivo.10,11

Prophylactic G-CSF with combination chemotherapy decreases the incidence of febrile neutropenia, documented infections, hospitalizations and antibiotic use.1,7-9 It allows the administration of a greater dose intensity of chemotherapy, as the number of dose reductions and dose delays because of neutropenia are decreased.1,7-9 NHL is a chemotherapy-sensitive tumour, and in vitro and clinical evidence indicates that tumour toxicity is strongly dose related.12,13 A study by Epelbaum and colleagues14 found that 5-year survival was significantly affected by the average relative dose intensity of the CHOP regimen (cyclophosphamide, doxorubicin, vincristine and prednisone) that patients with NHL received; 5-year survival was 66% and 20% respectively for patients receiving greater than or less than the median relative dose intensity (RDI) (p = 0.003). Bertini and colleagues15 demonstrated that, in elderly patients with aggressive NHL, prophylactic G-CSF increased the median RDI that patients received from 79% to 94%. The rates of remission and event-free survival in this study were both found to be significantly influenced by an RDI greater than 80% (remission: 89% v. 56%; event-free survival: 50% v. 18%; p = 0.05).15 A similar study by the South Western Oncology Group also demonstrated the importance of dose intensity on treatment outcome in NHL; it found that a 50% dose reduction in CHOP decreased event-free survival from 41% to 12%.16

Despite the decrease in neutropenia and febrile neutropenic events and the potential increase in dose intensity attainable with prophylactic G-CSF, objections to its use remain. The main objection lies in its cost-effectiveness. In the current study, we examined the incidence and severity of neutropenic complications -- including febrile neutropenic episodes, documented infections and chemotherapy dose reductions and dose delays -- in patients treated with combination chemotherapy for advanced-stage NHL at the University of Calgary. We then examined the potential clinical and economic implications of secondary prophylactic G-CSF, defined as G-CSF administered following a documented chemotherapy-induced neutropenic complication.

A retrospective chart review was performed on a consecutive sample of 42 patients with advanced-stage NHL who had been treated with a full course of systemic combination chemotherapy between Jan. 1, 1992, and Dec. 31, 1993. All charts were from the Tom Baker Cancer Centre, Calgary. Only patients between 18 and 70 years of age with biopsy-proven stage III, IV or IAE disease (using the Ann Arbor Staging System) were included (Table 1). Chemotherapy regimens varied by patient and the histological classification of the NHL and included CHOP (20 patients), VACOPB (7 patients), CNOP (6 patients), CVP (5 patients) and what we classified as "other" (1 patient each receiving VACOPBP, C-MOPP, CVP/CNOP and CHOP/ABVD) (Table 2). It is standard practice at our centre that all patients receiving systemic chemotherapy for NHL have blood counts done at least once per chemotherapy cycle. Fourteen patients were excluded from our study: patients treated for recurrent lymphoma, patients who had bone marrow transplants, patients who received radiation therapy, HIV-positive patients and patients who had already received prophylactic G-CSF.

The incidences of febrile neutropenic episodes, documented infections, hospitalizations and antibiotic use were recorded from the charts. Febrile neutropenia was defined as an oral temperature of greater than 38.58C with an ANC of less than 1000/?L.5 A documented infection was defined as a febrile illness with an obvious focus of infection. Neutropenia occurring as a nadir or at the time when the subsequent course of chemotherapy was begun was documented. Neutropenia was defined as an ANC of less than 1000/?L; severe neutropenia was defined as an ANC of less than 500/?L.5 Dose delays and reductions resulting from neutropenia were also recorded. A dose delay was defined as a delay of scheduled chemotherapy by at least 1 week. A dose reduction was defined as a reduction of at least 1 myelosuppressive drug by at least 10% of the intended dose.

Cumulative dose was determined for all patients based on standard chemotherapy regimen doses. Dose intensity was not calculated because of a number of confounding factors, including treatment of patients with a variety of chemotherapy regimens using a number of different agents, and dose modifications that occurred for reasons other than neutropenia.

The treatment costs (hospitalization and antibiotics) for neutropenic complications were obtained from the Foothills Hospital, Calgary.

This project was approved by the Conjoint Biomedical Research and Development Committee, the Foothills Ethics Committee and the Tom Baker Cancer Centre Research Committee.

Overall, 18 of the 42 patients (43%) experienced some form of neutropenic complication, including febrile neutropenic episodes, documented infections and chemotherapy dose modifications (Table 3). Eight patients (19%) had febrile neutropenic episodes; this incidence is lower than the 44% to 52% documented in the literature.1,17 In the CHOP group, 3 patients (15%) had febrile neutropenic episodes. Of all patients, 5 were treated with intravenously administered antibiotics for a total of 48 days, averaging 4.4 days of hospitalization per cycle. Three patients were treated with oral antibiotics as outpatients. Seven patients (17%), 5 of whom were treated with the CHOP regimen, had clinically documented infections while neutropenic, including upper respiratory tract infections, urinary tract infections, cellulitis and diarrhea. Five of these patients were treated with oral antibiotics, and received intravenously administered antibiotic therapy. One patient on the CHOP regimen died of pneumonia (mortality rate of 2.4%).

Forty-three percent of patients required dose modifications because of neutropenia - 10 patients required dose delays and 15 required dose reductions. Of the CHOP patients, 9 (45%) required dose modifications, including 6 patients who required dose delays and 7 patients who required dose reductions. In all but 4 patients, dose reductions were at least 25%. Of these 4 patients, 2 were receiving CHOP and had dose reductions of only 15% and 20%; however, they also experienced dose delays. Two patients receiving VACOPB also had dose reductions of only 20%; they did not require dose delays. Dose modifications for reasons other than neutropenia were not included in the analysis. As a result of dose modifications, 13 patients (31%) received less than 80% of the standard cumulative dose of myelosuppressive chemotherapy drugs because of neutropenia.

Seven patients (17%) became severely neutropenic, with an ANC of less than 500/?L at the nadir. Five of these patients were treated with CHOP. Three patients (7%), including 2 patients receiving VACOPB and 1 patient receiving VACOPBP, were neutropenic when their subsequent dose of chemotherapy was administered. Two of these patients were severely neutropenic at this time.

Analysis of neutropenia and neutropenic complications in these 42 patients showed that patients at high risk for neutropenic complications could be identified by at least one of the following criteria: (1) patients who became severely neutropenic, with an ANC nadir of less than 500/?L; (2) patients who were neutropenic, with an ANC nadir of less than 1000/?L, when their subsequent cycle of chemotherapy was administered; (3) patients who had experienced neutropenic complications in a previous cycle of chemotherapy.

By these criteria, 15 of the 42 patients (36%) were identified as being at high risk for neutropenic complications. Of these patients, 14 had neutropenic complications, which indicates a sensitivity of 77.8% for the criteria. Of the 27 remaining low-risk patients, only 4 had neutropenic complications, which indicates a specificity of 92.4% for the criteria. Of patients treated with CHOP, 8 (40%) were identified as being at high risk. In almost all cases, patients who were at high risk for neutropenic complications could be identified within the first 3 cycles of chemotherapy. These criteria are in agreement with those published in the literature.18-21

Glaspy and Jakway[22] found that the use of secondary prophylactic G-CSF, defined as prophylactic G-CSF administered only after the occurrence of a neutropenic complication in a previous cycle of chemotherapy, reduces by 50% the number of hospitalizations, the duration of hospitalization and antibiotic use in chemotherapy patients. Using the available published data, the potential economic implications of secondary prophylactic G-CSF were examined in our 15 high-risk patients.

The cost - including hospitalization, antibiotics and relevant investigations - of treating febrile neutropenic episodes and documented infections was calculated to be approximately $21 000 for the 15 high-risk patients and $8500 for the 8 high-risk CHOP patients (Table 4). The costs of treating febrile neutropenic episodes and documented infections in initial cycles, during which secondary prophylactic G-CSF would not have been given, have been excluded. With secondary prophylactic G-CSF, the costs of hospitalization and treatment of febrile neutropenia would have been reduced to $5500 in the total patient sample and $2700 in CHOP patients, based on a 50% reduction in the incidence of febrile neutropenic episodes, hospitalization and antibiotic use.22 The cost of secondary prophylactic G-CSF was determined to be $111 000 for the 15 high-risk patients and $46 500 for the 8 high-risk CHOP patients, based on a dosage of 5 ?g/kg per day at an average cost, including administration costs, of $150 per day.18,22,23 In each cycle, G-CSF therapy would have been started 24 hours after the administration of chemotherapy and continued until 24 hours before the next chemotherapy dose, or to a maximum of 10 days.18,22,23

Of the high-risk patients, 11 received less than 80% of the standard cumulative dose of chemotherapy. Had G-CSF been used, it was expected that only 1 of the 15 patients would have received less than 80% of the standard cumulative dose of chemotherapy.1 For CHOP patients, 4 received less than 80% of the standard cumulative dose of chemotherapy; it was expected that only 1 would have received less than 80% of the standard cumulative dose of chemotherapy had G-CSF been used.

Several previous studies have performed cost analyses of prophylactic G-CSF administration in patients receiving chemotherapy. As in our study, the researchers found that, although the incidence and treatment costs for infections and febrile neutropenic episodes are reduced, the cost of G-CSF greatly outweighs this reduction.24,25 These other studies, however, have failed to consider a possible increase in cure rates and survival rates resulting from the administration of an increased dose of chemotherapy with G-CSF support. Therefore, we performed a predictive incremental analysis in an attempt to include these factors in the analysis of the cost-effectiveness of G-CSF.26 This model has certain limitations, as it is based on several assumptions and on the limited published data available at present.

The cost analysis assumed a 41% survival rate for patients who received greater than 80% of the standard cumulative dose of chemotherapy.16 For patients receiving less than 80% of the standard cumulative dose of chemotherapy, the survival rate was assumed to be 12%.16 Standard cumulative dose of chemotherapy was used for the analysis rather than dose intensity because there are no comparable data in the literature for survival based on dose intensity. It was also assumed that life expectancy after cure is 10 years.16 Survival rates and assumed life expectancy were extrapolated from the South Western Oncology Group study, one of the few studies in the literature examining the effect of chemotherapy dose on survival in lymphoma.16

Using these assumptions, the expected life years of survival were calculated for the 15 high-risk patients with and without the use of secondary prophylactic G-CSF. This total was determined by multiplying the number of patients receiving more or less than 80% of the standard dose chemotherapy by the expected survival rate, then multiplying this figure by 10 years of life expectancy after survival. Without G-CSF, 4 patients would have received greater than 80% of the standard dose chemotherapy (expected survival rate, 41%), and 11 patients would have received less than 80% of the standard dose chemotherapy (expected survival rate, 12%). Cumulatively, the patients could be expected to survive 29.6 years. With prophylactic G-CSF, 14 patients would have received greater than 80% of the standard dose chemotherapy and only 1 less than 80% of the standard dose chemotherapy. These patients could therefore be expected to have a cumulative survival of 58.6 years.

The incremental cost per life year saved was determined by dividing the increased cost of G-CSF therapy by the improved outcome. The increased cost with the use of G-CSF equalled $93 300. Prophylactic G-CSF resulted in a 29-year increase in life expectancy for the 15 high-risk patients. Dividing these numbers results in a theoretical incremental cost per life year saved of $3300 through the administration of secondary prophylactic G-CSF. Performing the same analysis using the data from the CHOP patients results in a theoretical incremental cost per life year saved of $4700 through the use of secondary prophylactic G-CSF.

Comparing the use of secondary prophylactic G-CSF with other medical interventions, based on a similar analysis done by Smith, an incremental cost per life year saved of $3300 to $4700 is relatively low.27 For example, liver transplantation costs $237 000 per year for every life year it saves,27 captopril for hypertension costs $178 000 per life year saved, and autologous bone marrow transplantation for Hodgkin's disease costs $26 200 per life year saved.27

The incidence of neutropenic complications in the 42 patients with advanced-stage NHL was 43%. This included both chemotherapy dose modifications and episodes of febrile neutropenia and documented infections. Thirty-one percent of patients received less than 80% of the standard dose of chemotherapy because of neutropenic complications. We were unable to determine the survival rates in these patients, as many of the patients were lost to follow-up.

Analysis of the data showed that it was possible to identify patients who were at high risk for neutropenic complications. These patients could generally be identified within the first 3 cycles of chemotherapy. The criteria included (1) development of severe neutropenia, (2) neutropenia at the time of the next cycle of chemotherapy and (3) development of neutropenic complications in a previous cycle of chemotherapy. The sensitivity of these criteria was found to be 77.8% with a specificity of 92.4%. Fifteen of the 42 patients (36%) were determined to have been at high risk for neutropenic complications. Analysis of only CHOP patients brought similar results.

The theoretical incremental cost for the administration of secondary prophylactic G-CSF in high-risk patients was estimated to be $3300 per life year saved. The accuracy of this estimate is difficult to determine because of the limitations of the economic model used. Several assumptions, particularly the effect of dose intensity on survival rate in NHL and length of survival after cure, were necessary when applicable data were not available in the literature.

The effect of dose intensity on outcome in NHL is a question that remains unanswered in the literature. Few existing randomized trials directly examine the effect of dose intensity on outcome, and several studies that indirectly examined dose intensity did so by comparing different chemotherapy regimens with different drugs rather than by increasing the doses in a single regimen.28 The studies are also confounded by small sample sizes, short follow-up periods and relatively small increases in dose intensity, which may not be sufficient to produce a significant improvement in outcome. Studies have tended to focus on dose intensity, although it is not known if dose intensity is the critical factor or if the peak chemotherapy dose or cumulative dose is more important; however, improved cure rates with high-intensity chemotherapy with bone marrow transplant or stem cell rescue suggest that increased doses of chemotherapy are more efficacious. G-CSF is currently being used in a number of studies to decrease myelotoxicity and allow increased dose intensities; it is hoped that these studies will demonstrate whether dose intensity influences outcome.29,30

There is also a question about whether dose intensity has any prognostic significance for the patient. If a patient receives a lower dose intensity of chemotherapy because of neutropenic complications, is a less favourable outcome the result of decreased dose intensity or other prognostic factors? Are these the same factors that contribute to the patient's poor tolerance of chemotherapy? In a study by Lepage and colleagues,31 univariate and multivariate analyses of dose intensity and initial prognostic factors (increased lactate dehydrogenase level, poor performance status, advanced stage, high number of extranodal sites) demonstrated an absent or weak relation among these parameters, suggesting that patients who receive a decreased dose intensity of chemotherapy are not initially destined to have a poor outcome. Results of studies of dose intensity using prophylactic G-CSF may help to answer this question by allowing patients with a poorer prognosis to tolerate an increased dose intensity of chemotherapy.

While the estimation of the theoretical incremental cost of prophylactic G-CSF in this study may be limited by the assumptions used, the economic model developed does provide a method for considering both a decrease in morbidity and an increase in survival in the calculation of the cost-effectiveness of secondary prophylactic G-CSF. Such an economic model may provide physicians with a numerical value with which to justify the use of similar expensive therapies.

(Original manuscript received June 4, 1997; revised Dec. 23, 1997; accepted Jan. 8, 1998)

Clin Invest Med 1998;21(2):63-70.

Reprint requests to: Dr. Richard Charles Woodman, Department of Medicine, University of Calgary, 3330 Hospital Dr. NW, Calgary AB Neutropenic complications in non-Hodgkin's lymphoma

*Types of Stage IAE NHL included small bowel (2 patients), gastric, nasopharyngeal, scalp, thyroid and testicular. [+] CHOP = cyclophosphamide + doxorubicin + vincristine + prednisone. VACOPB = etoposide + doxorubicin + cyclophosphamide + vincristine [+] prednisone + bleomycin. CNOP = cyclophosphamide + mitanoxantrone + vincristine + prednisone. CVP = cyclophosphamide + vincristine + prednisone. [+]Other chemotherapy regimens included VACOPBP (VANCOPB + cisplatin), C-MOPP (cyclophosphamide + mechlorethamine + vincristine +procarbazine + prednisone), CVP/CNOP, CHOP/ABVD (ABVD = doxorubicin + bleomycin + vinblastine + dacarbazine).

[*]ANC = absolute neutrophil count. [dagger] = febrile neutropenic episodes.

*IV = intravenously administered. [+]CT = computed tomographic. [+]LFT = TO COME [+]CBC = complete blood count.