Immunotherapy and Radiation Therapy | BioOne

RADIATION RESEARCH

182, 252–257 (2014)

0033-7587/14 $15.00 Ó2014 by Radiation Research Society. All rights of reproduction in any form reserved. DOI: 10.1667/RR13707.1

Immunotherapy and Radiation Therapy: Considerations for Successfully Combining Radiation into the Paradigm of Immuno-Oncology Drug Development Elad Sharon,a,1 Mei-Yin Polley,b Michael B. Bernsteinc and Mansoor Ahmedd a Cancer Therapy Evaluation Program; b Biometric Research Branch and d Radiation Research Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute, National Institutes of Health, Bethesda, Maryland; and c Department of Radiation Oncology, Albert Einstein College of Medicine and Montefiore Medical Center, Bronx, New York

Sharon, E., Polley, M-Y., Bernstein, M. Immunotherapy and Radiation Therapy: Successfully Combining Radiation into Immuno-Oncology Drug Development. 252–257 (2014).

(Provenget, Dendreon) was approved in 2009 for patients with metastatic prostate cancer (1). Then in 2011, Ipilimumab (Yervoyt, Bristol-Myers Squibb) was the second agent to be approved for patients diagnosed with metastatic melanoma (2). While other anticancer agents, including chemotherapy, are certainly immunomodulatory, the mechanisms of action for these two agents uniquely herald a new era in the immune-mediated treatment of cancer. This trickle of approved cancer immunotherapy agents is likely to become a flood in the coming decade. Immunotherapy may have certain advantages over other targeted therapies, which emphasize its potential and applicability. Since immunotherapy often focuses on the tumor milieu (i.e., the tumor stroma and surrounding microenvironment) rather than the tumor itself, the spectrum of diseases that may respond to this modality is widen. As the recent success of anti-PD1/PD-L1 agents in non-small cell lung cancer and other malignancies demonstrates, tumors that are outside the scope of the ‘‘traditionally immunogenic tumors,’’ such as renal cell cancer and melanoma, can be successful targets of immunotherapy (3–5).

B. and Ahmed, M. Considerations for the Paradigm of Radiat. Res. 182,

As the immunotherapy of cancer comes of age, adding immunotherapeutic agents to radiation therapy has the potential to improve the outcomes for patients with a wide variety of malignancies. Despite the enormous potential of such combination therapy, laboratory data has been lacking and there is little guidance for pursuing novel treatment strategies. Animal models have significant limitation in combining radiation therapy with immunotherapy and some of the limitations of preclinical models are discussed in this article. In addition to the preclinical challenges, radiation therapy and immunotherapy combinations may have overlapping toxicities, and for both types of therapy, early and late manifestations of toxicity are possible. Given these risks, special attention should be given to the design of the specific Phase I clinical trial that is chosen. In this article, we describe several Phase I design possibilities that may be employed, including the 3 + 3 design (also known as the cohort of 3 design), the continual reassessment method (CRM), and the time-to-event continual reassessment method (TITE-CRM). Efficacy end points for further development of combination therapy must be based on multiple factors, including disease type, stage of disease, the setting of therapy and the goal of therapy. While the designs for future clinical trials will vary, it is clear that these two successful modalities of therapy can and should be combined for the benefit of cancer patients. Ó 2014 by Radiation Research Society

IMMUNOTHERAPY AND RADIATION

Combining immunotherapy with traditional methods of cancer treatment is an attractive concept. Radiation therapy is a fundamental component of cancer care and is utilized in curative and palliative settings. In addition to its tumoricidal capabilities, a recent review article described several preclinical and clinical studies showing the ability of radiation to activate the immune system and enhance immune responses (6). Interest in combining these two modalities was further accelerated with a recent unexpected finding reported in the New England Journal of Medicine. Postow et al. (7) reported a case of an abscopal effect, a phenomenon in which local radiotherapy is associated with the regression of metastatic cancer at a distance from the irradiated site, in a patient treated with ipilimumab for

INTRODUCTION

In the last decade, as targeted agents have dominated the world of oncology drug development, two agents have been approved that directly target the immune system as a means of treating cancer. The prostate cancer vaccine, sipuleucel-T 1 Address for correspondence: Cancer Therapy Evaluation Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute, National Institutes of Health, 9609 Medical Center Drive, Bethesda, MD 20892; e-mail: [email protected].

252

IMMUNOTHERAPY AND RADIATION THERAPY

melanoma. In addition to evidence of tumor shrinkage, the patient described was found to have antibody responses to the cancer-testis antigen NY-ESO-1, changes in peripheralblood immune cells, and increases in antibody responses to other antigens after radiotherapy. Given these findings, many have speculated that the abscopal effect initiated by local radiotherapy may be mediated by the activation and interplay with the immune system (7). This and other studies have shown that exposure to radiation can cause immunogenic modulation and induce an immune-mediated cell death (8). Thus, combinations with various immune targets, including immune checkpoint inhibitors, could further enhance the cytotoxic effects of radiation therapy. A recent analysis of patients with melanoma brain metastases suggested longer survival in patients treated with both radiosurgery and ipilimumab (9). Other retrospective analyses report similar data, with suggestions that the type and sequencing of radiation therapy and immunotherapy may have a major bearing on overall survival in treated patients (10). With this in mind, investigators have been undertaking efforts to combine immunotherapy with radiation therapy to take advantage of their potential synergistic effects. Numerous clinical trials are currently underway exploring the combination of radiation therapy with ipilimumab. As more immune checkpoint inhibitors and immune activators enter the clinic, the question of how best to combine these agents with radiation therapy becomes more critical. Currently, the U.S. National Cancer Institute’s Cancer Therapy Evaluation Program (CTEP) has a limited portfolio of trials exploring the use of radiation in combination with immunotherapeutic agents. The small number of open trials in part reflects limitations in preclinical models as well as other unique facets of trial design that fall outside the scope of the traditional paradigm. In contemplating future clinical trials, several considerations need to be addressed. PRECLINICAL MODELS

Prior to initiating a novel treatment regimen in humans, a solid foundation of preclinical data showing proof of concept is essential. Experimental animal models have long been employed to mimic human disease processes and provide guidance for future clinical trials. However, significant physiological and biochemical differences between animals and humans can limit the development of adequate animal models for translational research. For example, certain cell surface signaling molecules, which ultimately activate downstream signaling pathways, can differ between murine models and humans. In prototypical mouse models that study the autoimmune pathogenesis of systemic lupus erythematosus (SLE), animals are defective in Fas and FasL, which play critical roles in initiating apoptosis in T cells (11). In contrast, SLE patients rarely have this mutation (12, 13). Similarly, glucocorticoid-

253

induced TNFR-related (GITR) is a gene encoding for a member of the TNF receptor superfamily. Activation of GITR by GITRL plays a role in the activation of regulatory and effector T cells (14). Contrary to mouse GITR, human GITR is not expressed in peripheral CD4 þ and CD8þ effector cells. However, after its activation, both mouse and human GITR is expressed at high levels (15, 16). Along these lines, a recently published article by Seok et al. (17) detailed how genomic responses in mouse models poorly mimic human inflammatory diseases. To improve current models, they suggest that rigorous genomic descriptions to define the human disease could help guide animal models by reproducing diseases on a molecular basis rather than simply based on phenotype. Additionally, development of synthetic human models may grant the opportunity to better predict pathophysiology in patients. Despite these disparities, however, several preclinical studies have successfully shown the interplay between radiation therapy and immune modulators to help lay the groundwork for combination clinical trials. Some in vivo models have suggested that radiation therapy exerts its antitumor effects by activation of the immune system (18). In one study, investigators found that the therapeutic effects of radiation were limited to immunocompetent mice, indicating that a robust immune response was necessary for radiation necrosis (19). Several other studies have shown that an increase in anticancer immunity is present after radiation therapy (20–24). To explain these findings, one group proposed that radiation leads to enhanced tumor antigen presentation to the host immune system, thereby resulting in antitumor immunotherapy (25). Further, certain published reports have confirmed that radiation therapy can alter the immune marker milieu on the surface of cancer cells in vitro, making them better targets for immune-mediated cell kill (26, 27). Although these studies have eloquently described a partnership between radiation and immunotherapy, choosing consistently reliable animal models for oncologic trials provides specific challenges. There are several possible explanations for the discordance between these preclinical findings and the translation to clinical success, including difficulty in modeling the microenvironment and immune responses, the heterogeneous nature of solid tumors (28, 29), and the use of fast-growing cell lines (30). Furthermore, the tumors studied in animal models are small and exposed to radiation over a few days compared to large, radioresistant tumors frequently encountered in the clinic that are treated with a course of fractionated radiotherapy over several weeks. Therefore, the mechanisms involved in these relatively distinct settings to cause tumor regression may substantially differ from one another (31). In addition, preclinical studies commonly use transplantable tumor models, where inbred animals are inoculated with tumor cells derived from the same genetic strain. However, inbred animal models may not replicate the

254

SHARON ET AL.

inherent diversity in most human malignancies and are often immunologically deficient. An alternative to inoculating inbred animal models involves the utilization of genetically engineered mouse (GEM) models. These GEM models acquire tumors spontaneously with metastatic potential, which more accurately mimic the natural disease progression in humans and thereby better predict the efficacy of a novel treatment regimen in the clinic. However, GEM models may pose challenges similar to inbred animal models with inoculated tumor cells. While genetically engineered mice tend to be relatively immunocompetent, their spontaneously generated tumors have decreased mutational diversity when compared to most human tumors, particularly in the metastatic disease setting. TRANSLATION OF IMMUNOTHERAPY AND RADIATION THERAPY TO THE CLINIC

Preclinical data help provide the rationale for testing the combination of immunotherapy and radiation therapy in a clinical setting. However, other obstacles remain. First and foremost, as with any combination therapy, new regimens must be judged for safety and feasibility. Assessing toxicity in trials involving radiation therapy poses additional obstacles rarely seen with other oncologic treatments. Side effects from exposure to radiation occur in two phases, with early effects occurring within three months of completion of therapy and late effects that may arise years later. In contrast, other agents such as chemotherapy often cause toxicity within hours or days of treatment, allowing for dose modifications prior to instituting long-term sequelae. Furthermore, combining radiation therapy with immunotherapeutic agents may augment side effects, since these two modalities share a number of potential overlapping toxicities. While this is not a reason to avoid the combination altogether, it can give clinical trialists pause. For instance, radiation pneumonitis from thoracic radiotherapy and autoimmune pneumonitis caused by an immunotherapy agent could be a potentially fatal combination. Thus, the Phase I component of a clinical trial to determine safety of the regimen is of paramount importance. Once a regimen is judged to be safe, the next question is the effectiveness of the combination regimen. This seems to be a unique challenge in combining radiation and immunotherapy. The design of trials for radiation and immunotherapy combinations will be specific to tumor type, the stage of disease, the setting of therapy (i.e., neoadjuvant, adjuvant, concomitant) and the goal of therapy (i.e., palliative or curative). The main goal, even when using surrogate end points, is ultimately to ensure that patients experience benefit through an improvement in overall survival or an improvement in quality of life and a reduction of the cancer-related symptom burden, including the local control of disease.

STATISTICAL DESIGNS FOR PHASE I TRIALS WITH RADIATION AND IMMUNOTHERAPY COMBINATION

Like traditional Phase I trials, the primary objective of a Phase I radiation and immunotherapy trial is to determine the safe dose level of the novel treatment regimen. Referred to as the maximum tolerated dose (MTD), this represents the dose at which the percentage of patients experiencing the dose-limiting toxicity (DLT) is kept below a prespecified level. While it is often not explicitly stated, Phase I dose-finding studies are inherently associated with a fixed window within which the DLTs are expected to occur. Dose assignment decisions are then made based on DLTs observed within this time window, which is decided by the investigators at the design stage of the trial. For the estimated MTD to be relevant to the study drug, this window must realistically reflect the timeframe during which the DLTs are expected to occur. In practice, the most widely used dose-finding method in Phase I trials is the 3 þ 3, also known as the cohort of three design. This is a rule-based scheme in which the dose levels are fixed in advance and the dose escalation decision depends on the number of patients at a dose that experience a DLT (32–34). In recent years, many alternative designs to the standard 3 þ 3 design have been proposed, including the continual reassessment method (CRM) (35, 36) and its various extensions and modifications (37–41). The CRM has gained popularity since its introduction by O’Quigley in 1990 (35). It is often referred to as a model-based approach since it utilizes a mathematical model relating dose levels to the probability of DLT. The CRM utilizes information from all previously treated patients, updates the estimates of the DLT probability at each dose at each stage of the trial, and assigns a future patient adaptively to the dose estimated to be closest to the MTD (42). Since their inception, the statistical properties of the 3 þ 3 design and the CRM-based methods have been extensively studied (43–48). To date, the general consensus in the literature is that the 3 þ 3 design tends to be more conservative and thus, a ‘‘safer’’ protocol to follow, whereas the CRM tends to yield a more accurate estimate of the MTD with desirable statistical properties. Despite their differences, both designs may be feasible in trials where toxicities associated with the study drug occur fairly rapidly after drug administration. However, as described above, Phase I trials with radiation and immunotherapy could result in certain toxicities months to years after a course of therapy. The 3 þ 3 and CRM statistical designs are poorly suited in this setting since they require suspension of patient accrual while a given dose level is evaluated for DLTs. Dose decisions for the next patient (or next cohort of patients) are delayed until all of the patients treated at the current dose have been fully evaluated for DLT. Thus, when late toxicities are potentially expected, as in radiation and immunotherapy combination trials, these

255


designs may be impractical as they can result in excessive trial duration from frequent enrollment suspensions. Therefore, statistical designs accommodating late toxicity are warranted. A dose-finding design that does not require the trial to be closed to accrual while trial patients are observed for toxicities is the time-to-event continual reassessment method (TITE-CRM) (49). While the estimation procedure and the dose allocation rule of the TITE-CRM closely follow the classic CRM paradigm, the TITE-CRM design permits enrollment of new patients while previous patients remain under observation for DLTs. In effect, this model substantially reduces the trial duration compared to the CRM. In TITE-CRM, the dose decision for a new patient is facilitated by utilizing the full information from patients who have experienced a DLT and the partial information from patients who have been followed for a period of time free of DLT based on the length of their followup. Although the TITE-CRM offers a promising approach to incorporate both acute and late toxicities, it has some limitations. Like the CRM, the TITE-CRM design depends on the validity of various statistical modeling assumptions, including the length of the assumed DLT evaluation window and the distribution of time to DLT. When using this method, the basis of these assumptions relies on prior data and the understanding of potential treatment toxicities. Another notable shortcoming of the TITE-CRM is patient safety when rapid accrual is anticipated. Specifically, since toxicities are expected to be rare during the early stage of the trial, patients enrolled early are likely to contribute DLTfree information based on short follow-up. Since a TITECRM trial is always open to accrual, erroneously rapid dose escalation may occur early, exposing patients to potentially toxic doses. Polley (50) and Bekelle et al. (51) have proposed modifications to the TITE-CRM to address this issue. There is a common misconception in practice that innovative statistical designs are always better than the standard 3 þ 3 design. While continued research to develop and refine trial designs should be encouraged, it is equally important to recognize the added practical complexity newer designs bring about. Specifically, the design and execution of complex trial designs typically requires substantial resources and expertise. For example, protocols involving complex designs may require more time and effort to develop and justify to an institutional review board. In addition, when deciding the most suitable Phase I design to implement for a trial, it is imperative that both the statistician and the clinician understand the operating characteristics of the design under consideration and how these properties compare with the 3 þ 3 design in terms of patient safety and practical feasibility. Important parameters to consider may include percentage of patients expected to be treated at suboptimal or supra-optimal doses, expected trial duration and the likelihood that the trial ultimately correctly identifies the MTD. Such comparisons can only be

achieved with extensive statistical simulation work under the guidance of an experienced statistician. Furthermore, informatics and statistical software may not be readily available to implement newer designs. Even if a tool exists, any amendments or deviations from the original method may require additional work to tailor the software for each specific trial. For example, software for implementing the TITE-CRM design has been made available by the original TITE-CRM author (52) but in situations where aggressive dose assignment may be of concern, standard safety rules used in more traditional designs and additional design constraints to temper speedy dose escalation should be incorporated into the design to ensure patient safety. Implementation of these technical modifications will require a statistician with sound understanding of the method and proficient programming skills. Finally, complex designs are often associated with higher logistical costs than the standard 3 þ 3 design. For example, the sequential dose assignment decisions in a TITE-CRM trial are based on frequent updates of toxicity information gathered throughout the trial. These intra-trial adjustments require real-time data management and statistical oversight. In summary, no design is flawless, but several methods may prove useful. A team-based approach and a strong collaboration between clinicians and statisticians are critical to the success of a trial. Clinicians provide valuable insight into design elements such as dose levels of interest, relevant DLTs and acceptable DLT rates for the given disease process. The involvement of statisticians is crucial both at the design and execution stage of the trial. Ultimately, the choice of a Phase I study design should strike a reasonable balance between patient safety and the statistical rigor of the design. As any clinical investigator would agree, patient safety is of paramount importance. Thus, when the combination of immunotherapy and radiation therapy is anticipated to cause significant adverse events in a particular disease setting, a more conservative dose-escalation scheme should be adopted. Conversely, when minimal toxicity is expected, a more aggressive dose-escalation scheme may be appropriate. CONCLUSIONS

In spite of these challenges, radiation therapy and immunotherapy represent an attractive combination to translate further into clinical practice. Guidance from animal models will continue to help clinical trialists utilize resources in an efficient manner. In addition, careful considerations for statistical designs to appropriately accommodate the safety profiles of these combination regimens are vital to the success of future investigations. Once preclinical experiments and early phase clinical trials determine an effective approach to combining radiation therapy with immunotherapy, the full benefit of these regimens in human malignancies may be realized.

256

SHARON ET AL.

ACKNOWLEDGMENT The views expressed in this article are the personal opinions of the authors and do not necessarily reflect policy of the U.S. National Cancer Institute. Received: February 16, 2014; accepted: June 11, 2014; published online: July 8, 2014

REFERENCES 1. Kantoff PW, Higano CS, Shore ND, Berger ER, Small EJ, Penson DF, et al. Sipuleucel-T immunotherapy for castrationresistant prostate cancer. N Engl J Med 2010; 363:411–22. 2. Hodi FS, O’Day SJ, McDermott DF, Weber RW, Sosman JA, Haanen JB, et al. Improved survival with ipilimumab in patients with metastatic melanoma. N Engl J Med 2010; 363:711–23. 3. Brahmer JR, Drake CG, Wollner I, Powderly JD, Picus J, Sharfman WH, et al. Phase I study of single-agent antiprogrammed death-1 (MDX-1106) in refractory solid tumors: safety, clinical activity, pharmacodynamics, and immunologic correlates. J Clin Oncol 2010; 28:3167–75. 4. Topalian SL, Hodi FS, Brahmer JR, Gettinger SN, Smith DC, McDermott DF, et al. Safety, activity, and immune correlates of anti-PD-1 antibody in cancer. N Engl J Med 2012; 366:2443–54. 5. Brahmer JR, Tykodi SS, Chow LQ, Hwu WJ, Topalian SL, Hwu P, et al. Safety and activity of anti-PD-L1 antibody in patients with advanced cancer. N Engl J Med 2012; 366:2455–65. 6. Kwilas AR, Donahue RN, Bernstein MB, Hodge JW. In the field: exploiting the untapped potential of immunogenic modulation by radiation in combination with immunotherapy for the treatment of cancer. Front Oncol 2012; 2:104. 7. Postow MA, Callahan MK, Barker CA, Yamada Y, Yuan J, Kitano S, et al. Immunologic correlates of the abscopal effect in a patient with melanoma. N Engl J Med 2012; 366:925–31. 8. Gameiro SR, Jammeh ML, Wattenberg MM, Tsang KY, Ferrone S, Hodge JW. Radiation-induced immunogenic modulation of tumor enhances antigen processing and calreticulin exposure, resulting in enhanced T-cell killing. Oncotarget 2013; 5:403–16 9. Knisely JP, Yu JB, Flanigan J, Sznol M, Kluger HM, Chiang VL. Radiosurgery for melanoma brain metastases in the ipilimumab era and the possibility of longer survival. J Neurosurg 2012; 117:227–33. 10. Silk AW, Bassetti MF, West BT, Tsien CI, Lao CD. Ipilimumab and radiation therapy for melanoma brain metastases. Cancer Med 2013; 2:899–906. 11. Han B, Moore PA, Wu J, Luo H. Overexpression of human decoy receptor 3 in mice results in a systemic lupus erythematosus-like syndrome. Arthritis Rheum 2007; 56:3748– 58. 12. Mysler E, Bini P, Drappa J, Ramos P, Friedman SM, Krammer PH, et al. The apoptosis-1/Fas protein in human systemic lupus erythematosus. J Clin Invest 1994; 93:1029–34. 13. McNally J, Yoo DH, Drappa J, Chu JL, Yagita H, Friedman SM, et al. Fas ligand expression and function in systemic lupus erythematosus. J Immunol 1997; 159:4628–36. 14. Nocentini G, Ronchetti S, Petrillo MG, Riccardi C. Pharmacological modulation of GITRL/GITR system: therapeutic perspectives. Br J Pharmacol 2012; 165:2089–99. 15. Kober J, Leitner J, Klauser C, Woitek R, Majdic O, Stockl J, et al. The capacity of the TNF family members 4-1BBL, OX40L, CD70, GITRL, CD30L and LIGHT to costimulate human T cells. Eur J Immunol 2008; 38:2678–88. 16. Bianchini R, Bistoni O, Alunno A, Petrillo MG, Ronchetti S, Sportoletti P, et al. CD4( þ) CD25(low) GITR( þ) cells: a novel human CD4( þ) T-cell population with regulatory activity. Eur J Immunol 2011; 41:2269–78.

17. Seok J, Warren HS, Cuenca AG, Mindrinos MN, Baker HV, Xu W, et al. Genomic responses in mouse models poorly mimic human inflammatory diseases. Proc Natl Acad Sci U S A 2013; 110:3507–12. 18. Burnette BC, Liang H, Lee Y, Chlewicki L, Khodarev NN, Weichselbaum RR, et al. The efficacy of radiotherapy relies upon induction of type i interferon-dependent innate and adaptive immunity. Cancer Res 2011; 71:2488–96. 19. Lee Y, Auh SL, Wang Y, Burnette B, Meng Y, Beckett M, et al. Therapeutic effects of ablative radiation on local tumor require CD8 þ T cells: changing strategies for cancer treatment. Blood 2009; 114:589–95. 20. Blanquicett C, Saif MW, Buchsbaum DJ, Eloubeidi M, Vickers SM, Chhieng DC, et al. Antitumor efficacy of capecitabine and celecoxib in irradiated and lead-shielded, contralateral human BxPC-3 pancreatic cancer xenografts: clinical implications of abscopal effects. Clin Cancer Res 2005; 11:8773–81. 21. Camphausen K, Moses MA, Menard C, Sproull M, Beecken WD, Folkman J, et al. Radiation abscopal antitumor effect is mediated through p53. Cancer Res 2003; 63:1990–3. 22. Cotter SE, Dunn GP, Collins KM, Sahni D, Zukotynski KA, Hansen JL, et al. Abscopal effect in a patient with metastatic Merkel cell carcinoma following radiation therapy: potential role of induced antitumor immunity. Arch Dermatol 2011; 147:870– 2. 23. Shiraishi K, Ishiwata Y, Nakagawa K, Yokochi S, Taruki C, Akuta T, et al. Enhancement of antitumor radiation efficacy and consistent induction of the abscopal effect in mice by ECI301, an active variant of macrophage inflammatory protein-1alpha. Clin Cancer Res 2008; 14:1159–66. 24. Demaria S, Ng B, Devitt ML, Babb JS, Kawashima N, Liebes L, et al. Ionizing radiation inhibition of distant untreated tumors (abscopal effect) is immune mediated. Int J Radiat Oncol Biol Phys 2004; 58:862–70. 25. Reits EA, Hodge JW, Herberts CA, Groothuis TA, Chakraborty M, Wansley EK, et al. Radiation modulates the peptide repertoire, enhances MHC class I expression, and induces successful antitumor immunotherapy. J Exp Med 2006; 203:1259–71. 26. Garnett CT, Palena C, Chakraborty M, Tsang KY, Schlom J, Hodge JW. Sublethal irradiation of human tumor cells modulates phenotype resulting in enhanced killing by cytotoxic T lymphocytes. Cancer Res 2004; 64:7985–94. 27. Sharma A, Bode B, Wenger RH, Lehmann K, Sartori AA, Moch H, et al. gamma-Radiation promotes immunological recognition of cancer cells through increased expression of cancer-testis antigens in vitro and in vivo. PloS One 2011; 6:e28217. 28. Heppner GH. Tumor heterogeneity. Cancer Res 1984; 44:2259– 65. 29. Axelson H, Fredlund E, Ovenberger M, Landberg G, Pahlman S. Hypoxia-induced dedifferentiation of tumor cells–a mechanism behind heterogeneity and aggressiveness of solid tumors. Semin Cell Dev Biol 2005; 16:554–63. 30. Linder S, Shoshan MC. Is translational research compatible with preclinical publication strategies? Radiat Oncol 2006; 1:4. 31. Lawrence YR, Vikram B, Dignam JJ, Chakravarti A, Machtay M, Freidlin B, et al. NCI-RTOG translational program strategic guidelines for the early-stage development of radiosensitizers. J Nat Cancer Inst 2013; 105:11–24. 32. Lin Y, Shih WJ. Statistical properties of the traditional algorithmbased designs for phase I cancer clinical trials. Biostatistics 2001; 2:203–15. 33. Storer BE. Design and analysis of phase I clinical trials. Biometrics 1989; 45:925–37. 34. Ji Y, Wang SJ. Modified toxicity probability interval design: a safer and more reliable method than the 3 þ 3 design for practical phase I trials. J Clin Oncol 2013; 31:1785–91.


35. O’Quigley J, Pepe M, Fisher L. Continual reassessment method: a practical design for phase 1 clinical trials in cancer. Biometrics 1990; 46(1):33–48. 36. O’Quigley J, Shen LZ. Continual reassessment method: a likelihood approach. Biometrics 1996; 52(2):673–84. 37. Potter DM. Adaptive dose finding for phase I clinical trials of drugs used for chemotherapy of cancer. Stat Med 2002; 21:1805–23. 38. Heyd JM, Carlin BP. Adaptive design improvements in the continual reassessment method for phase I studies. Stat Med 1999; 18:1307–21. 39. Piantadosi S, Fisher JD, Grossman S. Practical implementation of a modified continual reassessment method for dose-finding trials. Cancer Chemother Pharmacol 1998; 41:429–36. 40. Braun TM. The bivariate continual reassessment method. extending the CRM to phase I trials of two competing outcomes. Control Clin Trials 2002; 23:240–56. 41. Goodman SN, Zahurak ML, Piantadosi S. Some practical improvements in the continual reassessment method for phase I studies. Stat Med 1995; 14:1149–61. 42. Garrett-Mayer E. The continual reassessment method for dosefinding studies: a tutorial. Clin Trials 2006; 3:57–71. 43. Rosenberger WF, Haines LM. Competing designs for phase I clinical trials: a review. Stat Med 2002; 21:2757–70. 44. O’Quigley J, Chevret S. Methods for dose finding studies in

45.

46. 47. 48.

49. 50.

51. 52.

257

cancer clinical trials: a review and results of a Monte Carlo study. Stat Med 1991; 10(11):1647–64. Chevret S. The continual reassessment method in cancer phase I clinical trials: a simulation study. Stat Med 1993; 12(12): 1093–108. Korn EL, Midthune D, Chen TT, Rubinstein LV, Christian MC, Simon RM. A comparison of two phase I trial designs. Stat Med 1994; 13(18):1799–806. Paoletti X, Kramar A. A comparison of model choices for the continual reassessment method in phase I cancer trials. Stat Med. 2009; 28(24):3012–28. Le Tourneau C, Lee JJ, Siu LL. Dose escalation methods in phase I cancer clinical trials. J Nat Cancer Inst 2009; 101:708– 20. Cheung YK, Chappell R. Sequential designs for phase I clinical trials with late-onset toxicities. Biometrics 2000; 56:1177–82. Polley MY. Practical modifications to the time-to-event continual reassessment method for phase I cancer trials with fast patient accrual and late-onset toxicities. Stat Med 2011; 30:2130–43. Bekele BN, Ji Y, Shen Y, Thall PF. Monitoring late-onset toxicities in phase I trials using predicted risks. Biostatistics. 2008; 9:442–57. Cheung YK. dfcrm: Dose-finding by the continual reassessment method. R package version 02-2. 2013: http://CRAN.R-project. org/package¼dfcrm.