Estimation of optimal dynamic treatment regimes

BACKGROUND

Recent advances in medical research suggest that the optimal treatment rules should be adaptive to patients over time. This has led to an increasing interest in studying dynamic treatment regime, a sequence of individualized treatment rules, one per stage of clinical intervention, which maps present patient information to a recommended treatment. There has been a recent surge of statistical work for estimating optimal dynamic treatment regimes from randomized and observational studies. The purpose of this article is to review recent methodological progress and applied issues associated with estimating optimal dynamic treatment regimes.

METHODS

We discuss sequential multiple assignment randomized trials, a clinical trial design used to study treatment sequences. We use a common estimator of an optimal dynamic treatment regime that applies to sequential multiple assignment randomized trials data as a platform to discuss several practical and methodological issues.

RESULTS

We provide a limited survey of practical issues associated with modeling sequential multiple assignment randomized trials data. We review some existing estimators of optimal dynamic treatment regimes and discuss practical issues associated with these methods including model building, missing data, statistical inference, and choosing an outcome when only non-responders are re-randomized. We mainly focus on the estimation and inference of dynamic treatment regimes using sequential multiple assignment randomized trials data. Dynamic treatment regimes can also be constructed from observational data, which may be easier to obtain in practice; however, care must be taken to account for potential confounding.

Reinforcement learning strategies for clinical trials in nonsmall cell lung cancer

Typical regimens for advanced metastatic stage IIIB/IV nonsmall cell lung cancer (NSCLC) consist of multiple lines of treatment. We present an adaptive reinforcement learning approach to discover optimal individualized treatment regimens from a specially designed clinical trial (a "clinical reinforcement trial") of an experimental treatment for patients with advanced NSCLC who have not been treated previously with systemic therapy. In addition to the complexity of the problem of selecting optimal compounds for first- and second-line treatments based on prognostic factors, another primary goal is to determine the optimal time to initiate second-line therapy, either immediately or delayed after induction therapy, yielding the longest overall survival time. A reinforcement learning method called Q-learning is utilized, which involves learning an optimal regimen from patient data generated from the clinical reinforcement trial. Approximating the Q-function with time-indexed parameters can be achieved by using a modification of support vector regression that can utilize censored data. Within this framework, a simulation study shows that the procedure can extract optimal regimens for two lines of treatment directly from clinical data without prior knowledge of the treatment effect mechanism. In addition, we demonstrate that the design reliably selects the best initial time for second-line therapy while taking into account the heterogeneity of NSCLC across patients.

Reinforcement learning design for cancer clinical trials

We develop reinforcement learning trials for discovering individualized treatment regimens for life-threatening diseases such as cancer. A temporal-difference learning method called Q-learning is utilized that involves learning an optimal policy from a single training set of finite longitudinal patient trajectories. Approximating the Q-function with time-indexed parameters can be achieved by using support vector regression or extremely randomized trees. Within this framework, we demonstrate that the procedure can extract optimal strategies directly from clinical data without relying on the identification of any accurate mathematical models, unlike approaches based on adaptive design. We show that reinforcement learning has tremendous potential in clinical research because it can select actions that improve outcomes by taking into account delayed effects even when the relationship between actions and outcomes is not fully known. To support our claims, the methodology's practical utility is illustrated in a simulation analysis. In the immediate future, we will apply this general strategy to studying and identifying new treatments for advanced metastatic stage IIIB/IV non-small cell lung cancer, which usually includes multiple lines of chemotherapy treatment. Moreover, there is significant potential of the proposed methodology for developing personalized treatment strategies in other cancers, in cystic fibrosis, and in other life-threatening diseases.

Randomized, placebo-controlled, phase II study of sequential erlotinib and chemotherapy as first-line treatment for advanced non-small-cell lung cancer

PURPOSE

This study investigated whether sequential administration of erlotinib and chemotherapy improves clinical outcomes versus chemotherapy alone in unselected, chemotherapy-naïve patients with advanced non-small-cell lung cancer (NSCLC).

PATIENTS AND METHODS

Previously untreated patients (n = 154) with stage IIIB or IV NSCLC and Eastern Cooperative Oncology Group performance status of 0 or 1 were randomly assigned to receive erlotinib (150 mg/d) or placebo on days 15 to 28 of a 4-week cycle that included gemcitabine (1,250 mg/m(2) days 1 and 8) and either cisplatin (75 mg/m(2) day 1) or carboplatin (5 x area under the serum concentration-time curve, day 1). The primary end point was nonprogression rate (NPR) at 8 weeks. Secondary end points included tumor response rate, NPR at 16 weeks, duration of response, progression-free survival (PFS), overall survival (OS), and safety.

RESULTS

The NPR at 8 weeks was 80.3% in the gemcitabine plus cisplatin or carboplatin (GC)-erlotinib arm (n = 76) and 76.9% in the GC-placebo arm (n = 78). At 16 weeks, the NPR was 64.5% for GC-erlotinib versus 53.8% for GC-placebo. The response rate was 35.5% for GC-erlotinib versus 24.4% for GC-placebo. PFS was significantly longer with GC-erlotinib than with GC-placebo (adjusted hazard ratio, 0.47; log-rank P = .0002; median, 29.4 v 23.4 weeks); this benefit was consistent across all clinical subgroups. There was no significant difference in OS. The addition of erlotinib to chemotherapy was well tolerated, with no increase in hematologic toxicity, and no treatment-related interstitial lung disease.

CONCLUSION

Sequential administration of erlotinib following gemcitabine/platinum chemotherapy led to a significant improvement in PFS. This treatment approach warrants further investigation in a phase III study.

Duration of Chemotherapy for Advanced Non–Small-Cell Lung Cancer: A Systematic Review and Meta-Analysis of Randomized Trials

Purpose

To determine if it is preferable to extend chemotherapy beyond a standard number of cycles in patients receiving first-line chemotherapy for advanced non–small-cell lung cancer.

Methods

We searched biomedical literature databases and conference proceedings for randomized controlled trials (RCTs) comparing a defined number of cycles with continuation of the same chemotherapy until disease progression, a larger defined number of cycles of identical chemotherapy, and a defined number of cycles of identical initial chemotherapy followed by additional cycles of an alternative chemotherapy. Meta-analysis was performed using the fixed effect model. The primary outcome was overall survival (OS); secondary outcomes included progression-free survival (PFS), adverse events (AE), and health-related quality of life (HRQL).

Results

We found 13 RCTs including 3,027 patients. Extending chemotherapy improved PFS substantially (hazard ratio [HR], 0.75; 95% CI, 0.69 to 0.81; P < .00001) and OS modestly (HR, 0.92; 95% CI, 0.86 to 0.99; P = .03). Subgroup analysis revealed that effects on PFS were greater for trials extending chemotherapy with third-generation regimens rather than older regimens (HR, 0.70 interaction v 0.92 interaction; P = .003). Extending chemotherapy was associated with more frequent AE in all trials where it was reported and impaired HRQL in two of seven trials.

Conclusion

Extending chemotherapy, particularly with a third-generation regimen, improved PFS substantially, but OS less so. Future trials should test extending treatment with more effective and/or better-tolerated agents.

Treatment paradigms for advanced stage non-small cell lung cancer in the era of multiple lines of therapy

The duration of first-line and the timing of second-line therapy for advanced non-small cell lung cancer has been an area of recent investigation. Five trials have been performed that have investigated shorter (3-4 cycles) versus longer duration of platinum-based therapy; four trials revealed an equivalent overall survival with the shorter duration of therapy, and one trial revealed superior survival with the longer duration of therapy. The toxicity and quality of life data has either been equivalent or favored the shorter duration of therapy. Two trials have investigated the timing of a second-line therapy after completion of four cycles of platinum-based therapy versus the standard treatment paradigm of initiating second-line therapy upon disease progression. Both of these trials have revealed a statistically significant improvement in the progression-free survival, and a trend towards improved survival for the earlier use of second-line therapy. Only 50 to 60% of patients on the standard treatment arm initiated second-line therapy, and the promising results observed are most likely related to the fact that a higher percentage of patients received second-line therapy on the experimental arm. Several trials have investigated maintenance chemotherapy, and these trials have not revealed a survival benefit probably due to the fact that many patients experience disease progression or unacceptable toxicity during the initial or maintenance therapy. The addition of a targeted agent (bevacizumab or cetuximab) to the initial chemotherapy and the continuation of the targeted agent after completion of the chemotherapy have yielded superior overall survival in comparison to chemotherapy alone. The incremental benefit of the maintenance therapy with the targeted agent is unknown.

Phase II study of pemetrexed and carboplatin plus bevacizumab with maintenance pemetrexed and bevacizumab as first-line therapy for nonsquamous non-small-cell lung cancer

PURPOSE

This study evaluated the efficacy and safety of pemetrexed, carboplatin, and bevacizumab followed by maintenance pemetrexed and bevacizumab in patients with chemotherapy-naive stage IIIB (effusion) or stage IV nonsquamous non-small-cell lung cancer (NSCLC).

PATIENTS AND METHODS

Patients received pemetrexed 500 mg/m(2), carboplatin area under the concentration-time curve of 6, and bevacizumab 15 mg/kg every 3 weeks for six cycles. For patients with response or stable disease, pemetrexed and bevacizumab were continued until disease progression or unacceptable toxicity.

RESULTS

Fifty patients were enrolled and received treatment. The median follow-up was 13.0 months, and the median number of treatment cycles was seven (range, one to 51). Thirty patients (60%) completed > or = six treatment cycles, and nine (18%) completed > or = 18 treatment cycles. Among the 49 patients assessable for response, the objective response rate was 55% (95% CI, 41% to 69%). Median progression-free and overall survival rates were 7.8 months (95% CI, 5.2 to 11.5 months) and 14.1 months (95% CI, 10.8 to 19.6 months), respectively. Grade 3/4 hematologic toxicity was modest-anemia (6%; 0), neutropenia (4%; 0), and thrombocytopenia (0; 8%). Grade 3/4 nonhematologic toxicities were proteinuria (2%; 0), venous thrombosis (4%; 2%), arterial thrombosis (2%; 0), fatigue (8%; 0), infection (8%; 2%), nephrotoxicity (2%; 0), and diverticulitis (6%; 2%). There were no grade 3 or greater hemorrhagic events or hypertension cases.

CONCLUSION

This regimen, involving a maintenance component, was associated with acceptable toxicity and relatively long survival in patients with advanced nonsquamous NSCLC. These results justify a phase III comparison against the standard-of-care in this patient population.

Second-line treatment of advanced non-small cell lung cancer

After failure of first-line chemotherapy for advanced non-small cell lung cancer, many patients remain candidates to receive further antitumor treatment. To guide clinical management of these patients and to suggest priorities for clinical research, an International Panel of Experts met in Naples (Italy) in April 2007. Results and evidence-based conclusions are presented in this article. Single-agent chemotherapy with docetaxel or pemetrexed is the recommended option for unselected patients with performance status 0 to 2 who are candidates for second-line chemotherapy for advanced non-small cell lung cancer. Docetaxel has demonstrated superiority compared with best supportive care. Pemetrexed has been shown to be noninferior to docetaxel, with a more favorable toxicity profile. Erlotinib is effective in pretreated patients, and can be given second-line in patients not suitable or intolerant to chemotherapy, and in all patients as third-line treatment after failure of second-line chemotherapy. Gefitinib failed to show superiority to placebo as second- or third-line treatment, but it has been shown to be noninferior to docetaxel. In selected patients such as lifetime nonsmokers or those of East-Asian ethnicity, erlotinib, or gefitinib (where licensed) may be considered as second-line treatment even if they are fit for chemotherapy. Best supportive care in addition to active treatment remains important for all patients, but may be the exclusive option for patients unsuitable for more aggressive therapy. Further research is mandatory, to find better treatments, and to identify clinical and molecular predictive markers of efficacy, both for chemotherapy and for novel biologic agents.

First line chemotherapy with planned sequential administration of gemcitabine followed by docetaxel in elderly advanced non-small-cell lung cancer patients: a multicenter phase II study

This multicenter phase II study evaluated, in chemonaive patients with stage IIIB–IV NSCLC, age ⩾70 and with a performance status 0–2, the activity, efficacy and tolerability of planned sequential administration of gemcitabine 1200 mg m⁻² on days 1 and 8 every 3 weeks for three courses followed by three cycles of docetaxel 37.5 mg m⁻² on days 1 and 8 every 3 weeks, provided there was no evidence of disease progression. A total of 56 patients entered the study. According to intention-to-treat analysis, the objective response rate was 16.0% (95% CI 7.6–28.3%); 23 patients (41.0%) had stable disease and 24 patients (43%) had progressive disease. Five patients who had a stable disease after three courses of gemcitabine obtained a conversion to partial response by docetaxel. Median time to progression was 4.8 months (95% CI 3.6–6.0 months) and median duration of survival was 8.0 months (95% CI 5.6–10.5 months). The 1-year survival rate was 34%. No grade 4 haematological toxicity was observed and grade 3 neutropenia and thrombocytopenia were reported in 5.4 and 3.6% of the patients, respectively. Grade 3/4 mucositis and grade 3 diarrhoea, both occurred in 3.6% of the patients and grade 3 asthenia was observed in 9% of patients. One patient reported a grade 4 skin toxicity. No treatment-related deaths occurred. Sequential gemcitabine and docetaxel is a well-tolerated and effective regimen in elderly advanced NSCLC patients.

Hot topics and landmark studies from the 43rd annual meeting of the American Society of Clinical Oncology

The results of several preclinical and clinical studies were reported by oncology professionals at the 43rd American Society of Clinical Oncology (ASCO) meeting, the largest international forum in which the latest achievements in cancer research are annually presented. The central theme this year was 'Translating Research into Practice', emphasizing the goal of forging stronger links between basic research and clinical practice. This review offers a critical, summarized selection of several of the foremost studies presented at the meeting. The focus is on the findings from randomized phase III trials that, in the authors' opinion, are most likely to have an immediate effect on clinical practice.

[Upcoming strategies in the treatment of non-small cell lung cancer stage IIIB/wet and IV]

Non-small cell lung cancer (NSCLC) is the most common malign lung tumour. It is diagnosed in the majority of patients at an advanced stage without a chance for cure despite substantial progress in the therapeutic armamentarium. The therapeutic goals in this situation are prolongation of life and symptom palliation. Both the classical chemotherapy as well as modern therapeutic strategies (tyrosinkinase inhibitors, bevacizumab) are able to achieve these goals. It is currently unclear how to incorporate the new strategies into first line therapy. At the ASCO congress 2007 numerous studies were presented for first line therapy of advanced NSCLC. A large phase III study with bevacizumab and a study with Docetaxel maintenance therapy are to be discussed in detail, followed by a short presentation of a study with platinum-free combinations, a new platinum combination with pemetrexed/carboplatin, the chemotherapy of patients with PS (2/3), and interesting studies with cetuximab.

[NSCLC: update on second line therapy following ASCO 2007]

Since the year 2000 second line therapy for pre-treated non-small cell lung cancer patients has been established. There are currently two chemotherapeutic agents - docetaxel (Taxotere) and pemetrexed (Alimta) - which have been approved for second line treatment in Austria in addition to the tyrosinkinase-inhibitor erlotinib (Tarceva). In randomised trials these agents have shown a clear advantage over best supportive care in pre-treated patients in terms of overall survival and quality of life. However, these compounds also cause considerable drug-specific toxicities. Therefore there is an urgent need for new treatment options with higher efficacy and a lower burden of toxicity. At ASCO 2007, results of randomised trials were presented concerning new chemotherapeutic agents, such as vinflunine as well as targeting agents such as gefitinib, without affecting outcome. Another trial comparing a standard dose of pemetrexed with a higher dose also showed no improvement in outcome. A comparison of immediate application with delayed application of docetaxel after primary treatment demonstrated improved progression-free survival in selected patients in the immediate study arm but this did not translate to a survival benefit. The results of these trials may add further treatment options to the present portfolio of agents and concepts in this setting and give some optimism for the near future. At present, no changes need to be made in the options for standard second-line treatment of NSCLC in view of the results presented at ASCO 2007.

Sequential Chemotherapy with Combination Irinotecan and Cisplatin Followed by Docetaxel for Treatment-Naïve Patients with Advanced Non-small Cell Lung Cancer

BACKGROUND

Sequential administration of platinum-based doublet therapy and then a taxane may reduce the risk of drug resistance and, therefore, improve treatment outcome. This study was designed to evaluate the efficacy and tolerability of sequential administration of irinotecan and cisplatin and then docetaxel in patients with advanced non-small cell lung cancer (NSCLC).

METHODS

Eligible patients received irinotecan in 60-mg/m2 infusions for 30 to 60 minutes on days 1, 8, and 15, and cisplatin in 75-mg/m2 infusions for 60 minutes on day 1 every 28 days for four cycles (IC). Regardless of the response, patients received up to four cycles of sequential docetaxel in 75-mg/m2 infusions for 60 minutes.

RESULTS

Forty-six patients with histologically confirmed chemotherapy-naïve stage IIIB or IV NSCLC were enrolled, of whom 42 were evaluable. The response rate at completion of chemotherapy with IC was 45.2% (95% confidence interval [CI]: 30.2%-60.3%). Five patients had improvement of disease status during sequential docetaxel, and seven patients had disease progression. Progression-free survival was 8.0 months (95% CI: 5.4-9.9 months), and the overall median survival was 14.6 months (95% CI: 9.8-17.9 months). The 1-, 2-, and 3-year survival rates were 54.3%, 22.6%, and 12.1%, respectively. The incidence of severe (> or =CTC V2 grade 3) neutropenia during IC was 23.9% compared with 95.7% for sequential docetaxel (p < 0.0001).

CONCLUSION

Sequential administration of IC and then docetaxel is feasible and is associated with a prolonged progression-free survival, but the current data do not confirm an improvement in treatment outcome by the sequential approach.