|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2000 American Society for Clinical Oncology Palliative Effect of Chemotherapy: Objective Tumor Response Is Associated With Symptom Improvement in Patients With Metastatic Breast CancerFrom the Catholic University of Nijmegen, Nijmegen, the Netherlands; National Cancer Institute of Canada Clinical Trials Group, Kingston; and Princess Margaret Hospital, Toronto, Ontario, Canada. Address reprint requests to Elizabeth Eisenhauer, MD, National Cancer Institute of Canada Clinical Trials Group, Queen’s University, 82-84 Barrie St, Kingston, Ontario, Canada K7L 3N6.
PURPOSE: Because one of the goals of chemotherapy for metastatic breast cancer is to provide symptom palliation, we were interested in identifying the relationship between tumor shrinkage and improvement in disease-related symptoms. PATIENTS AND METHODS: Three hundred patients enrolled onto a randomized trial of metastatic breast cancer formed the basis of our study. The nine most common baseline symptoms were identified and followed. Changes from baseline (improvement, stable, worsening) were defined using patient responses to a quality-of-life (QoL) questionnaire (the European Organization for Research and Treatment of Cancer EORTC QLQ-C30) as well as using graded toxicity data collected on case report forms (CRFs). The association between symptom improvement and tumor response was assessed using a linear trend test via a logistic regression model. RESULTS: The most commonly reported baseline symptoms were cancer pain in 38% (CRF data) and 81% of patients (QoL data) and tiredness in 26% (CRF data) and 89% (QoL data) of patients. Three symptoms—cancer pain, shortness of breath, and abnormal mood—showed a significant relationship between improvement and objective response, using both CRF and QoL assessments. Constipation, anorexia, and nausea showed a similar trend when QoL data were used but not when CRF information was used. The converse was seen for lethargy. There was no correlation between symptom change and response for cough and insomnia. CONCLUSION: For some symptoms, we found a significant association between symptom improvement and objective tumor regression. In these cases, symptom improvement was greatest in those patients who had complete or partial responses, followed by those with stable disease and then those with progressive disease. Further work in this area will be useful in determining the surrogate value of objective tumor response in identifying the efficacy of palliative chemotherapy.
BREAST CANCER IS the most common cancer diagnosed in women in Western countries. Many of these patients develop metastatic disease in the course of their illness. One of the goals of treating women with metastatic or recurrent breast cancer with chemotherapy is to provide symptom palliation. It is usually assumed that the number and severity of symptoms caused by the tumor burden will be decreased by chemotherapy and that there is a direct relationship between the tumor load and its symptomatic effect on the patient. If this is true, tumor response ought to be correlated with palliative benefit. The paucity of data on this subject is interesting, especially when it is compared with the amount of data on parameters such as objective tumor response, time to progression, and survival. However, a few studies have been conducted to assess the relationships between overall quality-of-life (QoL) scores, objective tumor response, and change in symptoms.1-6 Seidman et al explored the relationship between overall QoL scores and tumor response, as well as the predictive value of (baseline) QoL scores in patients with advanced breast cancer, but did not specifically examine tumor-related symptoms. In a group of patients with metastatic colorectal cancer treated with fluorouracil (5-FU) or 5-FU with leucovorin, a substantial number of patients were found to have reported improvement in their symptoms during the first 120 days of treatment.3 To our knowledge, objective tumor response (ie, complete response [CR] or partial response [PR]) has not been documented to be regularly associated with an improvement in disease-related symptoms over and above that seen in patients with disease stability or even disease progression. Therefore we were interested in using information from a large database of breast cancer patients registered on a trial of chemotherapy in metastatic disease to address two major questions:
Study and Data Collection The database for this study was that of the National Cancer Institute of Canada Clinical Trials Group MA-8 trial. In this trial, intravenous (IV) doxorubicin 60 mg/m2 day 1 every 3 weeks was compared with doxorubicin 40 mg/m2 IV day 1 plus vinorelbine 20 mg/m2 IV days 1 and 8 every 3 weeks in 303 women with breast cancer.7 To be eligible for the trial, patients had to have measurable or evaluable disease and have received one or no previous regimens of chemotherapy for recurrent/metastatic disease. Major trial end points included overall survival, QoL, response rate, and time to disease progression. To participate in the trial, patients had to complete QoL assessments. The QoL questionnaire used was the European Organization for Research and Treatment in Cancer Quality of Life Questionnaire-C30 (EORTC QLQ-C30). This questionnaire consists of 30 self-rating questions assessing global QoL, general symptoms (pain, fatigue, nausea, and vomiting), symptoms commonly reported by cancer patients (dyspnea, appetite loss, sleep disturbance, constipation, and diarrhea), the patient’s functional status (physical, role, cognitive, emotional, and social), as well as the perceived financial impact of the disease and its treatment.8 The questionnaire has a four-point category response scale: 1, not at all; 2, a little bit; 3, quite a bit; and 4, very much. Thus the higher the score, the more severe the symptom. The EORTC QLQ-C30 questionnaire is well-validated.8,9 Included in the questionnaire was a checklist specifically designed for patients with metastatic breast cancer who were receiving chemotherapy, the 13-item Mammary Cancer Checklist (National Cancer Institute of Canada Clinical Trials Group, unpublished data; questionnaire validated by D. Osoba and D. Warr, personal communication, October 1999). The baseline questionnaire had to be completed within 7 days before randomization. On study, a second questionnaire was to be completed on day 1 of cycle 3 before treatment was given, and thereafter, questionnaires were to be completed after every third cycle. A French translation of the questionnaire was available for francophone patients. Disease-related symptoms, as well as treatment related effects, were recorded on CRFs and graded according to National Cancer Institute of Canada Clinical Trials Group expanded Common Toxicity Criteria. Nurses and data managers completed the trial-specific case report forms at baseline and every 3 weeks (ie, at the first day of each new cycle). These forms captured baseline symptoms graded by severity and were to record changes in grade on each subsequent report. Toxicity grades ranged from 0 (absent) to 4 (very severe or life-threatening). A physical examination was performed every cycle. In addition, if the initial radiologic study results were positive, radiologic studies were repeated before the fourth cycle, and, if a complete response (CR) or partial response (PR) was documented, they were repeated after a further two cycles. The schedule of investigations for this study is shown in Fig 1.
Selection of Patients and Symptoms To Be Followed for Change on Treatment All patients eligible for the trial were included in this study. Using the baseline CRF data, the nine most common symptoms present at baseline were identified. All were thought to be either directly or indirectly related to malignant disease. These symptoms were followed in those patients in whom the symptom was present at baseline and who had at least one follow-up CRF submitted while they underwent treatment. Ten QoL questions addressing the same set of nine symptoms were also identified. For the symptom "mood," the two questions that best expressed abnormal mood (worry and depression) were selected, and for the other eight symptoms, one matching question was identified. Patients with a given symptom present at baseline were then observed for change, provided they had at least one follow-up questionnaire submitted in which there was an answer to the appropriate question. The relationship between symptom change and best objective response for both the CRF group and the QoL group was carried out in those patients who were assessable for objective tumor response.
Definitions for Symptom Responses Improvement.
Worsening.
Stable.
Definitions for objective tumor response.
For both measurable and evaluable disease, two observations made not less than 4 weeks apart were required.
Comparison of Symptom Response and Tumor Response For each patient with a given symptom, their symptom response while they were on treatment was determined (improvement, stable, worsening) using the definitions given earlier. These outcomes were then correlated with best objective tumor response in 3 x 3 tables (improvement, stable, and worsening/CR + PR, SD, and PD). Thus for each of the identified symptoms, two 3 x 3 tables were generated: the first showing the relationship between symptom change and objective response, using the QoL results, and the second showing the same relationship using CRF recorded symptom data. Results by randomized arm were not considered separately because the efficacy results of the two treatment arms were highly similar (response rate and median response duration for the doxorubicin arm were 30.5% and 6.8 months, respectively, and those for the doxorubicin and vinorelbine arm were 38% and 7.2 months, respectively).7 Logistic regression was used to test the effect of best tumor response on the probability of the patients’ having a symptom response. The probability of symptom response was modeled as an outcome of tumor response, where tumor response was defined as an ordinal variable with the values -1 for progression, 0 for stable, and +1 for improvement. Therefore, it was necessary to convert the 3 x 3 tables into 3 x 2 tables by grouping together patients who were stable or in whom a given symptom had worsened. Statistical analyses were done using the logistic procedure of the SAS computer software program (version 6.12).10
Patient Population and Common Baseline Symptoms Of the 303 patients randomized onto the study, 300 were eligible. Baseline and at least one follow-up CRF were submitted for all 300 patients. A baseline QoL questionnaire was obtained for 284 patients. Patients did not always answer all questions on the baseline as well as on the follow-up questionnaires. The nine most common symptoms present at baseline were cancer pain, constipation, lethargy/fatigue, shortness of breath, cough, nausea, mood change, anorexia, and insomnia. Table 1 lists these symptoms and their frequency by grade, according to baseline CRF data. Table 2 lists the 10 matching QoL questions, along with the numbers of patients who had indicated that the symptom described was present at baseline. In the last column of each table, the number of patients who were assessable for analysis for each symptom is recorded. Patients were considered to be assessable for the analysis if they had both an objective response recorded and a follow-up CRF (Table 1) or QoL questionnaire with the appropriate question answered (Table 2). Depending on the symptom and the method of evaluation, 88% to 96% of all patients with a specific baseline symptom were assessable and included in the analysis of that symptom. By definition, the CRF and QoL symptoms identified for follow-up were the same, but for each baseline symptom, when Tables 1 and 2 are compared, there were more patients identified as having the symptom when the QoL questionnaire was used. The most common baseline symptom recorded on the CRFs was cancer pain (115 patients [38%]) and that on the QoL questionnaires was tiredness (245 patients [89%]).
Objective Tumor Response As listed in Table 3, 289 patients were assessable for objective response. In the 145 patients assessable on the doxorubicin and vinorelbine arm, the overall response rate was 38% (seven CRs and 48 PRs). On the doxorubicin single-agent arm, the response rate was 30.5% (five CRs and 39 PRs in 144 patients). These differences were not significant (P = .2, Fisher’s exact test).
Symptom Response Table 4 depicts the percentages of patients having symptom improvement, stability, or worsening for each of the nine baseline symptoms as assessed by both the CRF data and the QoL questionnaires. Overall, when the CRF data are used, between 32.4% and 77.1% of patients experienced improvement in a given symptom. Anorexia was the most likely to improve (77% of patients) and lethargy the least (32.4%) likely to improve, using our definitions. The range was similar when the QoL data were examined: 33% to 70.8% of patients showed symptom improvement, depending on the specific symptom examined. Interestingly, in the QoL data, the symptoms most and least likely to show improvement were nausea and lethargy, respectively.
Cancer pain. As assessed by CRF data, of the 111 assessable patients with cancer pain at baseline, 67 (60.4%) improved, 31 (27.9%) remained stable, and only 13 (11.7%) worsened while on treatment. The distribution over the three objective tumor response categories showed that patients with a CR or PR were far more likely to have an improvement in pain (84.9% improved; Table 5). In contrast, 61.0% of the patients with SD and 15.8% of the patients with PD experienced an improvement in their pain, as defined for this study (P < .01).
Analysis of change in cancer pain, using QoL data, demonstrated highly similar results. Overall, 62.7% of the patients had improvement, 24.5% remained stable, and 12.7% worsened. Relating QoL response to objective tumor response showed the same pattern as the CRF data, although this was not as dramatic (Table 5). Improvement in pain was seen in 73.7% of the CR/PR patients, 63.2% of the patients with SD, and in 22.7% of the patients who had PD. This observed trend was also statistically significant (P < .01). Constipation. Constipation was identified at baseline in 74 assessable patients on CRFs and in 117 assessable patients in QoL questionnaires. Of the 74 patients, 31 (41.9%) had an improvement in this symptom while undergoing treatment, 35 (47.3%) remained stable, and eight (10.8%) worsened. The groups of patients with objective CR/PR and SD had almost identical rates of improvement in constipation (Table 5), and there was no significant relationship between the probability of improvement in constipation and objective response. The symptomatic response rate for constipation for the patients followed up with regard to their QoL questionnaires was 72 (61.5%) of 117. Twenty-seven patients (23.1%) were stable, and 18 patients (15.4%) had worsening constipation. In contrast to the CRF data, logistic regression showed a significant relationship between likelihood of improvement in this symptom, as assessed by QoL and objective tumor response (P = .027). Lethargy/tiredness. Out of the 71 assessable patients with baseline lethargy on CRFs, 23 (32.4%) improved, 27 (38.0%) remained stable, and 21 (29.6%) worsened. Forty-four percent of the patients with a CR or PR had improvement in lethargy. Of the patients with SD, 29.4% had improvement in the symptom. Only 10.0% of the patients with PD showed improvement in lethargy. This trend was significant (P = .046). Of the patients followed by questionnaires, 74 (33.0%) out of 224 had symptomatic improvement, 85 (37.9%) remained stable, and 65 (29.0%) worsened. The distribution of symptom responders according to the three objective tumor response categories was not significantly different (P = .428). Shortness of breath. Of the 66 assessable patients with shortness of breath recorded on the baseline CRFs, 36 (54.5%) showed improvement, 23 (34.8%) were stable, and seven (10.6%) worsened. The numbers of patients with a symptom response in the CR/PR and SD category were 17 (68.0%) and 19 (55.9%), respectively. None of the patients followed up using CRFs and who had objective disease progression had an improvement of their shortness of breath. This relationship between symptom improvement and objective response was significant (P = < .01). When the data from the QoL questionnaires were assessed, 85 (53.5%) of the patients had improvement in shortness of breath, 65 (40.9%) remained stable, and only nine (5.7%) worsened. As is depicted in Table 5, the distribution pattern over the three objective response categories demonstrates that patients with SD are less likely to have an improvement in shortness of breath, compared with the patients with a CR or PR. As well, patients with PD are less likely to have an improvement in this symptom, compared with the patients with SD. Once more, this trend was statistically significant (P = .016) Cough. Cough was reported in 57 assessable patients on the baseline CRFs and in 140 patients on the baseline QoL questionnaires. Of the group of 57 patients, 28 (49.1%) had improvement in cough, 23 (40.4%) were stable, and six (10.5%) had worsening. Of the patients who reported cough on their baseline questionnaire, 126 were assessable. Of these 126 patients, 82 (65.1%) showed improvement, 22 (17.5%) were stable, and another 22 (17.5%) worsened while on study. The proportion of patients followed by CRF whose cough improved was similar for objective CR/PR patients and SD patients (approximately 50% each; Table 5). The patients followed using the QoL questionnaire showed a similar pattern: 68.8% of the CR/PR category had improvement in cough, as did 67.7% of the patients with SD. In neither case was there a significant relationship between improvement in cough and tumor response. Anorexia/appetite. As recorded on the CRFs, 35 assessable patients had anorexia present on baseline CRF and 136 assessable patients reported loss of appetite on the baseline questionnaires. Twenty-seven (77.1%) of the patients followed up using the CRFs showed an improvement in anorexia, six patients (17.1%) were stable, and two patients (5.7%) showed worsening. The distribution over the three objective tumor response categories showed that a higher proportion of patients (91.7%) with a CR or PR had improvement in anorexia, compared with 72.2% of the patients with SD and 60.0% of the patients with PD. None of the patients with CR or PR had worsening of their anorexia. However, the P value for this observation does not reach significance. The QoL data showed a similar pattern: overall, 66.2% had improvement, 24.3% remained stable, and 9.6% worsened. The distribution also showed the same pattern: in the patients with CR or PR, 82.2% improved, as did 62.5% of the patients with SD, and 42.1% of the patients with PD. In contrast to the CRF group, the relationship between QoL reported symptoms and tumor response was significant (P < .01). Mood/worry and depression. Forty-three assessable patients presented with abnormal mood on the baseline CRFs. The majority of these 43 patients (22 patients [51.2%]) remained stable while on treatment. Fifteen (34.9%) showed improvement in mood, and six (14.0%) worsened. There was a trend which showed that patients with tumor response had a higher rate of improvement in mood (52.6%) than did those with either stable disease or PD (P = .019). The outcome for the two QoL questions regarding worry and depression showed 126 patients (58.6%) and 108 patients (60.3%) with improvement, respectively. Seventy patients (32.6%) remained stable with regard to the question about whether they felt worried the past 7 days, and 8.8% worsened. For the question that addressed depression, 47 patients (26.3%) were stable and 24 (13.4%) worsened. Both of these questions demonstrated a positive relationship between likelihood of improvement and objective tumor response (P = .012 and .046, respectively) (Table 5). Nausea. In total, 44 patients had nausea documented on the baseline CRF. Of these 44, 30 (68.2%) had improvement, seven (15.9%) remained stable, and seven (15.9%) worsened while on study. There was no relationship between tumor response and improvement in nausea (61.5%, 75%, and 57.1% improved in patients with CR/PR, SD, and PD, respectively; P = .965). None of the patients in the CRF group with PD had worsening of nausea (not shown in tables). The QoL assessment shows an improvement in nausea for 80 patients (70.8%). Twenty-seven (23.9%) were stable, and only six (5.3%) worsened. Eighty-one percent of the patients with CR/PR had an improvement in nausea, as assessed by the QoL questionnaire. Of the patients with SD, 72.6% improved. In contrast to the CRF results, the QoL results demonstrate a significant relationship (P < .01) between tumor response and likelihood of improvement in nausea. Review of the QoL data demonstrated that there were no patients in the CR/PR group with a worsening of their nausea (not shown in tables). Insomnia. Only a small number (23) of assessable patients had insomnia recorded on the baseline CRFs. Ten (43.5%) of them improved, 13 (56.5%) remained stable, and no patients reported worsening insomnia. Six of the patients with improvement in insomnia were in the CR/PR category. The remaining four patients all had SD as their best-ever tumor response status. On the QoL questionnaires, 175 assessable patients reported insomnia. Of these 175, 97 (55.4%) reported improvement of their insomnia, 49 (28.0%) didn’t experience any change, and 29 (16.6%) had more trouble sleeping. Fifty-six percent of the patients with a CR/PR had improvement in insomnia, as did 61.1% of the patients with SD and 23.5% of patients with PD. These findings were not significant.
We examined the relationship between objective tumor response and symptom improvement in a group of 303 patients enrolled onto a trial of chemotherapy in metastatic breast cancer. This was a retrospective analysis that used CRF and QoL data collected while the patients were on treatment. To our knowledge, this type of analysis, which focused on tumor-related symptoms, has not been carried out before. The nine most common symptoms identified and evaluated were cancer pain, constipation, lethargy, shortness of breath, cough, anorexia, nausea, abnormal mood (worry/depression), and insomnia. By intent, the same symptoms were evaluated by both CRF and QoL assessments. For any given symptom, the QoL questionnaires identified a much higher number of patients with that symptom. This observation is in keeping with those of many others who have noted that patient self-assessments reveal more numerous symptoms (whether due to treatment or disease) than do observer assessments of the same patients.11-15 We recognized that this might also be the case in our study sample but felt it would still be valuable to determine whether the pattern of change of symptoms in relationship to tumor shrinkage was the same, regardless of the source of symptom information. Ranking symptoms by their relative frequencies for both the CRF group and the QoL group showed some differences between the two groups. However, pain and fatigue were among the three most common symptoms in the CRF group as well as in the QoL group. Overall, anorexia and nausea demonstrated the highest rates of symptom improvement in both the CRF and QoL groups. It is plausible that these symptoms were all either directly or indirectly related to the presence of metastatic disease. Cancer pain, nausea, anorexia, fatigue, and shortness of breath are examples of symptoms that could be directly related to tumor burden. Constipation, insomnia, and worry are more likely to be indirect manifestations of the presence of disease either related to the knowledge of its presence (insomnia, worry) or secondary to narcotic analgesics (constipation). It was our hypothesis that if cancer-related symptoms improve in some patients who receive "palliative" chemotherapy, then the probability of improvement may be related to objective change in tumor size. That is, if symptoms are related to tumor burden, patients with CR/PR ought to have a greater likelihood of symptom improvement than would those with SD, who in turn should have higher rates of improvement than would those with PD. To assess this, we arbitrarily defined changes in symptoms to reflect states of "improvement" and "worsening" as well as "stable" (no change) for both CRF (toxicity-type) data and QoL data. In both cases, we felt that an improvement could reasonably be argued to have occurred if there was a decrease by one category (or level) on the scale of measurement. A patient had to achieve the improvement status on only one occasion while on therapy to qualify as a symptom responder. We specifically did not require improvement to be maintained for a minimum period because we were cognizant of the bias this requirement could introduce to our intended comparisons with objective response. If "improvements" had to be maintained for the period of two QoL questionnaires, for example, then the only patients who could improve would be those who received six cycles of chemotherapy, something that only CR or PR patients would be likely to do. This would bias any attempt to correlate objective response with symptom improvement because by definition only responders would be on trial long enough to meet symptom-improvement criteria. In fact, by using this definition, several, but not all, symptoms showed the hypothesized hierarchical relationship between improvement in symptoms and objective response category: responding patients having the highest rates of symptom improvement followed by those with SD and then those with PD. A significant relationship of this type was seen when both CRF and QoL data for cancer pain, shortness of breath, and abnormal mood were used (Fig 2). For constipation, nausea, and anorexia, a significant trend was shown only when QoL data were analyzed (Fig 3). In contrast, lethargy showed this pattern of behavior only when the CRF data were analyzed but not when QoL information was used. Cough and insomnia did not show any trend at all (Fig 4).
Nausea was one of the symptoms that showed a discrepancy between the CRF and QoL assessments. Interpretation of this finding must take into consideration that patients who had nausea at baseline may experience nausea while on study from two different sources: disease and chemotherapy. Usually chemotherapy-induced nausea will start after the administration of the drug and continue for only a few days. The intent of the CRF data is to capture all events that occurred during the cycle. This is in contrast to the QoL assessment, in which patients were asked about their state during the 7 days immediately before completion of the questionnaire (ie, at the end of each cycle). These different windows of assessment might explain the differences in behavior between the CRF and QoL assessments of nausea and their relationships to tumor response. Challenging this explanation, however, is the fact that the majority of patients with baseline nausea improved, according to our CRF-based criteria. The changes in lethargy/fatigue were interesting. Although the CRF data showed a clear pattern of increasing likelihood of improvement in this symptom with tumor response, the QoL data were unimpressive. To further address this, we examined the QoL data in the subset of patients who had lethargy recorded on CRFs (Table 6 and Fig 5). The pattern in this subset of QoL data was now similar to that of the CRF data. We can only speculate on the reasons for this, but one may be that the toxicity criteria definitions of lethargy used for CRF completion relate it to functional performance status, whereas the QoL question is much more vague: "Have you been tired in the past week?" Thus the QoL responses would capture not only those with true lethargy meeting toxicity criteria definitions but also those who had a normal functional status but who were sleep-deprived or fatigued without having an effect on the activity level. The assessment of this symptom is difficult and needs further study because many researchers have shown fatigue to be a major concern to patients.
Although the information provided by this study is quite useful in extending the understanding of the relationship between objective tumor response and symptom change, it is obvious that this study has its limitations. Although all data used in this study were collected prospectively, the fact that we decided to look for these relationships after completion of the trial meant that the data set was "fixed." As a result, it was not possible to retrieve missing data (ie, incomplete questionnaires), which slightly decreased the number of assessable patients. A prospective design would have offered the possibility of relating objective tumor response and symptom change at certain moments in time. Another limitation of this study related to our inability to relate symptom changes to lesser degrees of tumor shrinkage (ie, "minor responses"). Although our observations would have been strengthened by showing a level of symptom improvement in minor responders intermediate between true complete/partial responders and true no-change patients, the database did not contain this measurement end point. Comparing time-dependent variables might bias the results such that patients who have SD or a CR/PR will receive more cycles of treatment, compared with those who have PD. Thus these patients would have a greater chance of having at least one CRF or one QoL questionnaire showing an improvement in their symptoms while they were on treatment, sufficient to meet our arbitrary definitions of a symptom responder. However, if this alone were at play, all symptoms that were followed should have shown the same pattern of correlation with objective tumor response, which was not the case. As noted previously, relatively few studies have addressed the question of the effect of palliative chemotherapy on change in (overall) QoL or in disease-related symptoms. Because the intent of palliative chemotherapy in metastatic disease is to offer symptom relief, this small number of studies is quite surprising. Furthermore, many of these studies have looked at a summary score of QoL rather than individual symptoms. Although assessing global QoL impact is important, examining only global scores may obscure important and significant changes in disease-related symptoms. In a recent publication, Bernhard et al16 documented that certain parameters in QoL assessments were more likely to show higher scores at certain time points on therapy in breast cancer patients who responded to second-line endocrine therapy versus scores in patients with SD or disease progression. In that report, definitions of improvement in symptoms were not provided. Rather, differences between groups in linear analog scores were assessed at different points in time (correcting for baseline values). In a study in patients with metastatic colorectal carcinoma, Sullivan et al3 identified pain and tiredness (asthenia) as the two most common symptoms. Utilizing the Functional Living Index–Cancer, they did not find overall QoL improved in concert with tumor response. In a small study by Tester et al2 in patients with non–small-cell carcinoma, an attempt was made to correlate objective response with subjective changes in QoL. Only for patients with PD was a relationship found (ie, worsening in QoL was more common in patients with PD). This is in contrast to our findings, in which objective response was associated with a greater probability of symptom improvement for some of the symptoms. An explanation might be the small number of patients in the Tester et al study and the use of another QoL questionnaire, the Functional Assessment of Cancer Therapy (FACT) questionnaire. In a large randomized trial in colorectal cancer, Poon et al6 examined several modulated 5-FU regimens and documented wide variability in objective response rates (10% to 43%). Interestingly, the proportion of patients in each arm with symptom improvement on at least one occasion was proportional to the objective response rate in that arm. Thus 34% of patients had symptom improvement in the arm with a 10% response rate, compared with 69% of patients with symptom improvement in the arm with a 43% objective response rate. These data suggest that in the setting of palliative chemotherapy for colorectal cancer, symptom improvement may occur in parallel with objective tumor regression. Other trials have also addressed the issue of symptom palliation, although not necessarily in relation to tumor regression. In a randomized trial of mitoxantrone and prednisone versus prednisone alone in prostatic carcinoma, Tannock et al17 showed that there was a benefit to the combination arm in terms of symptom (ie, pain) improvement. Because objective response is difficult to document in this disease, they explored whether a fall in the serum marker prostate-specific antigen (PSA) was correlated with pain relief. From their article, it can be calculated that 20 (58%) of 53 patients with a fall in PSA of 25% or more from baseline had a symptom response, as opposed to only seven (12%) of 58 patients without a fall in PSA. Doyle et al18 assessed the palliative impact of chemotherapy in 27 ovarian cancer patients who had relapsed, using the EORTC QLQ-C30 and FACT-O questionnaires. They documented improvement in global and emotional function in 50% to 70% of patients but did not specifically examine individual symptoms and the pattern of their change with respect to objective response. Our results agree with the findings of others in that we have shown that symptoms of metastatic disease show improvement in some patients during the course of palliative chemotherapy. We further found that for many baseline disease-related symptoms, there is an association between the likelihood of improvement on chemotherapy and objective tumor response. Thus objective response might be a surrogate for symptom improvement in some cases. Additional prospective studies addressing the relationship between palliative and objective response outcomes will be useful in determining the utility of either outcome in predicting the other.
Supported in part by the National Cancer Institute of Canada, the Stichting Nijmeegs Universitair Fonds, and Glaxo-Wellcome Inc. We thank the following individuals for their assistance and helpful comments in completing this project: Prof Dr D.J.Th. Wagener, University of Nijmegen, Nijmegen, the Netherlands; David Osoba, MD, QOL Consulting, Vancouver, British Columbia, Canada; and Susan Marlin, National Cancer Institute of Canada Clinical Trials Group.
This study was performed as part of the scientific internship requirements of the doctor of medicine program at the University of Nijmegen for one of the coauthors (P.G.) in the National Cancer Institute of Canada Clinical Trials Group.
1. Seidman AD, Portenoy R, Yao TJ, et al: Quality of life in phase II trials: A study of methodology and predictive value in patients with advanced breast cancer treated with paclitaxel plus granulocyte colony stimulating factor. J Natl Cancer Inst 87:1316-1322, 1995 2. Tester WJ, Jin PY, Reardon DH, et al: Phase II study of patients with metastatic non-small cell carcinoma of the lung treated with paclitaxel by 3-hour infusion. Cancer 79:724-729, 1997[Medline] 3. Sullivan BA, McKinnis R, Laufman LR: Quality of life in patients with metastatic colorectal cancer receiving chemotherapy: A randomized, double-blind trial comparing 5-FU versus 5-FU with leucovorin. Pharmacotherapy 15:600-607, 1995[Medline] 4. Osoba D, Murray N, Gelmon K, et al: Quality of life, appetite, and weight change in patients receiving dose-intensive chemotherapy. Oncology 8:61-64, 1994[Medline] 5. Ellis PA, Smith IE, Hardy JR, et al: Symptom relief with MVP (mitomycin C, vinblastine and cisplatin) chemotherapy in advanced non-small-cell lung cancer. Br J Cancer 71:366-370, 1995[Medline] 6. Poon MA, O’Connell MJ, Moertel CG, et al: Biochemical modulation of fluorouracil: Evidence of significant improvement in survival and quality of life in patients with advanced colorectal carcinoma. J Clin Oncol 7:1407-1417, 1989[Abstract] 7. Norris B, Pritchard K, James K, et al: A phase III comparative study of vinorelbine (VNB) combined with doxorubicin (DOX) versus doxorubicin alone in metastatic/recurrent breast cancer (MBC): A National Cancer Institute of Canada (NCIC CTG) study. Proc Am Soc Clin Oncol 15:98a, 1996 (abstr 59)
8.
Aaronson NK, Amedzai S, Bergman B, et al: The European Organization for Research and Treatment of Cancer QLQ-C30: A quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst 85:365-76, 1993 9. Niezgoda HE, Pater JL: A validation of domains of the core EORTC Quality of Life questionnaire. Qual Life Res 2:319-325, 1993[Medline] 10. SAS Institute Inc:SAS Language Guide, Release 6.03 edition. Cary, NC,SAS Institute, 1988 11. Paul N, Pater JL, Whitehead M, et al: Methods of toxicity data collection: An evaluation of the relative effectiveness of the case report flow sheet, the patient symptom diary and the quality of life questionnaire. Controlled Clin Trials 12:648, 1991 (abstr A79) 12. Macquart-Moulin G, Viens P, Bouscary ML, et al: Discordance between physicians’ estimations and breast cancer patients’ self-assessment of side-effects of chemotherapy: An issue for quality of care. Br J Cancer 76:1640-1645, 1997[Medline] 13. Justice AC, Rebnek L, Hays RD, et al: Sensitivity, specificity, reliability, and clinical validity of provider-reported symptoms: A comparison with self-reported symptoms. J Acquir Immune Defic Syndr Hum Retrovirol 21:126-133, 1999 14. Youngblood M, Williams PD, Eyles H, et al: A comparison of two methods of assessing cancer therapy-related symptoms. Cancer Nurs 17:37-44, 1994[Medline] 15. Parliament MB, Danjoux CE, Clayton T: Is cancer treatment toxicity accurately reported? Int J Radiat Oncol Biol Phys 11:603-608, 1985[Medline]
16.
Bernhard J, Thurlimann B, Hsu Schmitz S-F, et al: Defining clinical benefit in postmenopausal patients with breast cancer under second-line endocrine treatment: Does quality of life matter? J Clin Oncol 17:1672-1679, 1999
17.
Tannock IF, Osoba D, Stockler MR, et al: Chemotherapy with mitoxantrone plus prednisone or prednisone alone for symptomatic hormone-resistant prostate cancer: A Canadian randomized trial with palliative end points. J Clin Oncol 14:1756-1764, 1996 18. Doyle C, Pintilie M, Crump M, et al: Do outcomes meet expectations in patients undergoing palliative chemotherapy in advanced ovarian cancer? Proc Am Soc Clin Oncol 16:413a, 1997 (abstr 1475) Submitted June 9, 1999; accepted March 1, 2000.
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2000 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
, CR/PR; , SD; 
, PD.
