Reflections on medical oncology: an appeal for better clinical trials and improved reporting of their results

Approach to the treatment of multiple myeloma: a clash of philosophies

In this Perspective, we summarize some of the most contentious issues surrounding diagnosis and treatment of myeloma. We outline how a fundamental clash of philosophies, cure versus control, may be at the heart of many of the controversies. From the very definition of the disease to risk stratification to the validity of current clinical trial endpoints, we highlight the major areas of debate and provide alternative viewpoints that have implications for trial design and interpretation, as well as clinical practice.

Measuring response in a post-RECIST world: from black and white to shades of grey

The unprecedented pace of therapeutic development in oncology has created a climate in which the traditional methods of evaluating agent activity might no longer be adequate. How is the field transitioning to new endpoints in early drug development and what are the difficulties in this transition? Here, we will explore the historical context for the current criteria for tumour response evaluation and some of the pitfalls in using these standards when testing newer anticancer agents for activity. We will argue that the current drug development environment dictates different outcome measurements and therefore more imaginative and rigorous early-phase trial designs.

Bone metastasis: review and critical analysis of random allocation trials of local field treatment

PURPOSE

Compare and contrast reports of random allocation clinical trials of local field radiation therapy of metastases to bone to determine the techniques producing the best results (frequency, magnitude, and duration of benefit), and relate these to the goals of complete relief of pain and prevention of disability for the remaining life of the patient.

METHODS AND MATERIALS

Review all published reports of random allocation clinical trials, and perform a systematic analysis of the processes and outcomes of the several trial reports.

RESULTS

All trials were performed on selected populations of patients with symptomatic metastases and most studies included widely diverse groups with regard to: (a) site of primary tumor, (b) location, extent, size, and nature of metastases, (c) duration of survival after treatment All trial reports lack sufficient detail for full and complete analysis. Much collected information is not now available for reanalysis and many important data sets were apparently never collected. Several of the variations in patient and tumor characteristics were found to be much more important than treatment dose in the outcome results. Treatment planning and delivery techniques were unsophisticated and probably resulted in a systematic delivery of less than the assigned dose to some metastases. In general the use and benefit of retreatment was greater in those patients who initially received lower doses but the basis and dose of retreatment was not documented. Follow-up of patients was varied with a large proportion of surviving patients lost to follow-up in several studies. The greatest difference in the reports is the method of calculation of results. The applicability of Kaplan-Meier actuarial analysis, censoring the lost and dead patients, as used in studies with loss to follow-up of a large number of patients is questionable. The censoring involved is "informative" (the processes of loss relate to the outcome) and not acceptable since it results in artificial elevation of the frequency of response. Overall, higher dose fractionated treatment regimens produced a better frequency, magnitude, and duration of response than lower dose single-fraction regimens. Relapse after initial response was frequent. The "median duration of relief" was much shorter than the "median duration of survival" post-treatment. Thus the "net pain relief" is far less than the goal of pain relief for the total duration of life after treatment.

CONCLUSIONS

The pain relief obtained in all studies is poor and our care practices need to be improved. Many patients never achieved complete relief and for most who did, the duration of relief was much less than their period of survival after treatment. Higher dose, fractionated treatments produced a greater frequency, magnitude, and duration of response with an improved "net pain relief." Additional trials with selection of comparable cases, good definition of extent of disease, exemplary treatment, and complete follow-up are required.

Response rate as an endpoint for evaluating new cytotoxic agents in phase II trials of non-small-cell lung cancer

BACKGROUND

Response rate (RR) has been used as a defining endpoint of new-agent phase II trials for non-small-cell lung cancer (NSCLC). However, tumor responses to chemotherapy do not always result in prolonged survival of patients with this disease.

DESIGN

Single-agent phase II trials were identified by a MEDLINE search of the period from 1976 to 1995. Associations between RR, median survival time (MST) and characteristics of patients who entered the trial, including tumor extent, performance status and prior chemotherapy, were studied by using the logistic regression model.

RESULTS

A total of 183 treatment arms in 176 trials (including 10 randomized phase II trials) were identified. The overall RR in the 6768 evaluable patients was 11%. Eleven drugs, cisplatin, epirubicin, ifosfamide, edatrexate, irinotecan, vinorelbine, docetaxel, paclitaxel, etoposide, vindesine, and 254-S, produced a RR of more than 20%. An MST of eight months or longer was obtained with 12 drugs, but there were cases in which no objective responses were produced by these drugs. MST was correlated with RR (r = 0.504, P < 0.0001), but ranged broadly at a given level of RR. Multiple linear regression analysis showed a significant correlation between RR and MST (regression coefficient = 0.60, P = 0.00003) after adjustment for other variables.

CONCLUSIONS

RR was significantly correlated with MST in single-agent phase II trials for NSCLC, but there is room for further consideration of the endpoint of these trials.

Analysis and presentation. Clinical trials on prostate cancer

OBJECTIVE

To present guidelines for the analysis and presentation of clinical trials on prostate cancer.

METHODS

Textbooks in statistics and oncology were searched for information, as were separate articles on the topic. Previously published advice was fused with own experience.

RESULTS

Minimum key points are given for the sections: Introduction, Materials and Methods, Results, Discussion, and Summary. The importance of 1 primary question in any clinical trial is stressed. The value of a detailed presentation of the trial design, the patient population and the inclusion/exclusion criteria, the characterization of the disease, the treatment schedules, and toxicity is underlined. Application of various statistical methods for different endpoints is suggested. Maturity of data, time for publication, and avoidance of publication bias are discussed. Some common pitfalls in the statistical analyses of clinical results are indicated. The impact of prognostic factors, proper staging procedures, and secondary treatments on the interpretation of survival analysis is pointed out. A shift from the (mis-)use of the P value in favor of confidence intervals is strongly encouraged. The use of comparing the survival of responders versus nonresponders is to be abandoned. A few practical hints concerning the presentation are offered. The minimum of data that should be presented in absolute numbers is indicated. Also, the data that should be provided in both graphic and numeric format are exemplified. Examples of essential graphic illustrations are provided. The need for improvements in the design analysis, and presentation of clinical trials is reemphasized. Finally, numerous references are listed. The article is addressed not only to authors and readers of clinical trials, but also to editors of medical journals.

CONCLUSION

The suggested guidelines may be useful in the analysis, presentation, and interpretation of clinical trials on prostate cancer. Moreover, compliance with these guidelines may facilitate comparisons with other similar trials and also, the incorporation of single studies into metaanalyses.

Cognitive impairment and quality of life in long-term survivors of malignant brain tumors

Thirtysix long-term survivors following the treatment of a malignant supratentorial brain tumor were examined for cognitive functions and global level of autonomy. Eighteen patients were symptom-free (SF) and 18 had clinical and neuroradiological recurrence (RE). The control group included 30 healthy subjects. All subjects underwent a neuropsychological battery for general and specific cognitive functions. The level of autonomy was assessed by means of the Karnofsky Performance Scale (KPS) for oncological patients. SF patients showed less impairment than RE patients both at the tests, as well as on the KPS. The cognitive deficits were subclinical in most SF patients, the tests for attention, memory and word fluency being the most sensitive in detecting subtle dysfunctions. The association between tumor location and specific cognitive deficits was inconstant in both patient groups. The results suggest that even subtle cognitive deficits can prevent SF long-term survivors from returning to premorbid autonomy and occupations, and that neuropsychological tests may be used as complementary routine indicators of their quality of life. Furthermore, our data show that, in selected patients, combined treatments and therapeutic insistence do not necessarily have the same deleterious effects.

Quality of life self-reports from 200 brain tumor patients: comparisons with Karnofsky performance scores

To test a method of assessing quality of life, 200 primary brain tumor patients in an outpatient clinic answered a 20-minute questionnaire covering ten aspects of quality of life. These results were compared with Karnofsky performance scale (KPS) scores, taking age into account. Among patients with KPS 90-100 (two-thirds of the patients), the KPS alone was difficult to interpret. The questionnaire, with its specific questions related to the key dimension of well-being, provided a more definitive assessment of status. The central importance of well-being was supported by its strong statistical relationships with other dimensions. Particularly, well-being was related to freedom from depression (p < 0.0001), active social life (p < 0.0001), energy (p < 0.01), and fewer symptoms (p < 0.05). The KPS was more useful in differentiating the other one-third of the patients (KPS 50-80) but was highly sensitive to age. KPS scores, therefore, may have been unreliable. Depression, reported by half of the patients, was not predicted by the KPS when age was excluded from the regression analysis but was related to the scores in well-being and socializing. Neither depression, well-being, nor socializing was influenced by age. Thus, the questionnaire directly assessed these central, emotionally based variables, particularly among patients with satisfactory KPS. Such an assessment is especially crucial as a supplement to the KPS in evaluating brain tumor survivors, whose emotional well-being is often severely challenged by treatment after surgical excision.

Neoadjuvant treatment of stages T2 to T4 bladder cancer with cis-platinum, cyclophosphamide and doxorubicin

In an ongoing phase II study 17 patients with potentially operable transitional cell carcinoma of the bladder (stages T2 to T4, Nx, Mo) have been treated with intravenous cis-platinum (50 mg.per m.2), cyclophosphamide (400 mg.per m.2) and doxorubicin (40 mg.per m.2). They were to receive 3 treatments at 3-week intervals before cystectomy and 2 treatments at 3-week intervals commencing 5 weeks after cystectomy. Of 17 patients 14 (82 per cent) completed all 3 preoperative treatments but only 7 (41 per cent) continued on to complete the entire 5 treatments. In most cases incomplete therapy was due to patient refusal. Toxicity was low as measured by World Health Organization standards. Of the 17 patients 9 (53 per cent) exhibited objective tumor response (pathological downstaging or greater than 50 per cent reduction of tumor volume determined by either computerized tomography scan and/or endoscopic examination. When the determination was made by endoscopy the changes were dramatic and not borderline.) No patient demonstrated a pathological complete response. All 9 of the responders (100 per cent) remain clinically free of disease at a median follow-up of 19 months (range 4 to 30 months). The 8 nonresponders have done poorly with 5 dead of disease, 1 alive with pelvic recurrence and 2 free of disease at 4 and 12 months. These tumor response rates compare favorably with other cis-platinum-based combination regimens. The response to the chemotherapy appears to be an important prognostic indicator. Phase III trials must be conducted to determine whether this neoadjuvant chemotherapy regimen has a significant effect on long-term patient survival.

Chemotherapy of urothelial tract tumors

Recent data from Phase II trials in patients with advanced transitional cell carcinoma of the urothelial tract suggest combination chemotherapy regimens are inducing a higher number of complete remissions (CR), and an overall response rate between 50% and 70%. Most active combination regimens are cisplatin + methotrexate based or cisplatin + Adriamycin (doxorubicin) based. As single agents, cisplatin has a response rate of 30% in 320 patients, methotrexate, 29% in 236 cases, and Adriamycin, 17% in 248 cases. With each drug used singly, however, complete response is uncommon. Other active single agents include vinblastine (16% in 38 cases) and mitomycin C (13% in 42 cases). New agents being evaluated which show some promise include gallium nitrate, carboplatinum, and other antifols. In a trial by the Northern California Oncology Group which evaluated a combination of cisplatin, methotrexate, and vinblastine (CMV), 28% of 50 cases achieved a CR lasting 44 weeks, and 28% a partial remission (PR) sustained for 29 weeks. A limited number of cases required surgical debulking for obtainment of CR status. At the University of Michigan, a trial of cisplatin and dichloromethotrexate induced responses in over 60% of cases. The regimen of methotrexate, vinblastine, Adriamycin, and cisplatin (M-VAC) has been reported to induce CR in 37% of cases, and PR in an additional 31%. In the latter trial at Memorial Hospital in over 100 cases with bidimensionally measurable advanced disease, the median survival of CR has not yet been reached at 28 months, whereas those who achieve PR survive 12 months versus 6 months for nonresponders. Indirectly, the success of such combination regimens is apparent from the increasing number of central nervous system relapses, without systemic recurrence, in complete responders. Additional data indicate that cisplatin + methotrexate, without the addition of other drugs, is also an active regimen. The attainment of CR in 20% to 40% of cases given these multidrug regimens has led to adjuvant and neoadjuvant protocols. Although results of randomized prospective trials have not yet been reported, preliminary Phase II data are promising.

The design and conduct of clinical trials in childhood brain tumors

The orderly conduct of clinical research trials in pediatric oncology has laid the foundation for modern curative therapy in several childhood malignancies. Such types of large clinical trials are relatively new to pediatric neuro-oncology. New treatments such as single and combination chemotherapy agents are studied in a Phase II trial in children with specific recurrent primary brain tumors. Computerized tomography (CT) and magnetic resonance imaging (MRI) scans and myelography allow the accurate determination of objective responses. The relative lack of promising Phase II trials has hampered the design of new Phase III studies for a variety of primary childhood brain tumors. The results of several Phase II trials may be incorporated into a prospective randomized Phase III study in which the endpoint is disease-free survival rather than response rate. Because of the rarity of specific types of primary childhood brain tumors, randomized Phase III trials are conducted more easily by cooperative cancer study groups. This article suggests some guidelines for the conduct of these Phase II and III trials in children with primary brain tumors so that accurate data may be accrued in an efficient manner.

13-Cis-retinoic acid: pharmacology, toxicology, and clinical applications for the prevention and treatment of human cancer

Retinoids, particularly 13-cis-retinoic acid, have shown great promise against a number of benign, but serious dermatological conditions. In animal models, 13-cis-retinoic acid functions is a potent antipromoter whether a cancer has been initiated by chemical, physical, or viral agents. Additionally, substantial antiproliferative activity of this compound has been demonstrated in vitro in many culture systems. Clinical activity noted against several types of skin malignancies has led to several investigations to determine the anticancer activity of 13-cis-retinoic acid. Response of a variety of preneoplastic and neoplastic lesions of epithelial histology has been demonstrated. The toxicity of 13-cis-retinoic acid largely reflects its tissue distribution with skin and subcutaneous side-effects limiting dose escalation. The pharmacology and pharmacokinetics of 13-cis-retinoic acid has been explored in a number of patients and a long terminal half-life demonstrated. This review will discuss 13-cis-retinoic acid as a good model for a biological response modifier.

Thymic peptide hormones: basic properties and clinical applications in cancer

The manuscript will provide an in-depth and critical review of the nomenclature, biochemistry, biological properties, and a summary of published and on-going clinical trials with all reported thymic preparations, including both partially purified thymic factors (e.g., thymosin fraction 5, thymostimulin) as well as purified and synthesized thymic peptides (e.g., thymosin alpha 1, thymulin). Particular emphasis will be placed on which thymic peptides should be categorized as true hormones. In addition, the comparative biochemistry and biological activity in animals will be summarized and contrasted for all the currently available thymic factors. The effects, in vitro of thymic factors, on peripheral blood lymphocytes isolated from normal donors and patients with primary immunodeficiency disorders, autoimmune disorders, and neoplastic disorders will also be reviewed. Finally, a detailed critical summary of the clinical trials performed with each of the thymic preparations will be presented with an emphasis on treatment of patients with cancer.

Vindesine in the treatment of metastatic breast cancer

Thirty-five patients with advanced metastatic breast cancer refractory to prior chemotherapy were treated with vindesine given at a fixed dose as a continuous 5-day infusion of 1.5 mg/day every 4 weeks. All patients were considered evaluable, and there were four patients with partial responses for more than 3 months (11%) and 13 patients with stable disease (37%). Two of the four responders had had disease progression on other vinca alkaloids. None of the responders had proven doxorubicin resistance. Side-effects included myelosuppression, neurotoxicity, nausea, stomatitis and fever, but these were seldom dose-limiting. The results--together with the results of other single-agent studies of vindesine summarized in the paper--indicate that the drug is an active agent in advanced breast cancer. However, the optimum way of administering vindesine and its inclusion in first-line therapy needs further study.

Cyclophosphamide, adriamycin, and 24-hour infusion of cis-diamminedichloroplatinum (II) in the management of patients with advanced head and neck neoplasms

Twenty-eight patients with measurable or evaluable, regionally advanced or metastatic head and neck cancer received the combination of cyclophosphamide (C), adriamycin (ADR), and a 24-hour infusion of cis-diamminedichloroplatinum (II) (P). Most patients had received extensive prior surgery and radiation therapy, but only two had prior chemotherapy. We observed a 46% response rate (13/28) which included five complete responders and eight partial responders. Nine of the 13 patients responded within the initial month of treatment. The median response duration for the 13 responding patients was 7.5 months. Moderate to severe nausea and vomiting, and alopecia were the most significant toxicities. Myelosuppression (WBC less than 4,100 cells/mm3) occurred in 90% of patients but there were no episodes of sepsis, nor did we detect any meaningful impairment in renal function.