Li Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Li Liu has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
Nolan A. Wages
Department of Biostatistics Virginia Commonwealth University Richmond Virginia USA
2 shared publicationsPavel Mozgunov
MRC Biostatistics Unit, University of Cambridge, Cambridge, UK
2 shared publications- Weishi Chen
MRC Biostatistics Unit University of Cambridge Cambridge UK
2 shared publications - Graeme F. Murray
Huntington Health, An Affiliate of Cedars‐Sinai Huntington Hospital Pasadena California USA
1 shared publication Hyun Lee
Virginia Commonwealth University Massey Comprehensive Cancer Center Richmond Virginia USA
1 shared publication- Keri Renee Maher
Virginia Commonwealth University Massey Comprehensive Cancer Center Richmond Virginia USA
1 shared publication
Latest Publications
- 30 September 2026
- Published by Wiley in Pharmaceutical Statistics
- Vol. 25 (6) , e70120
TITE ‐ LOCRM12 : A Local Continual Reassessment Method for Drug Combination Optimization Based on Late‐Onset Toxicity and Efficacy Outcomes
The primary goal of Phase I/II combination trials is to identify the optimal biological dose combination (OBDC), defined as the dose combination that achieves the highest efficacy while maintaining an acceptable toxicity rate. Several designs have been proposed for such trials. Among them, the local continual reassessment method for Phase I/II trials (LOCRM12) performs well when compared with other model‐based methods in most of the scenarios studied. However, LOCRM12 assumes that both toxicity and efficacy outcomes are fully observed before enrolling the next cohort, an assumption that may not hold in real‐world settings, where late‐onset outcomes are increasingly common with the emergence of novel therapies such as molecularly targeted agents and immunotherapies. To address this limitation, we propose a time‐to‐event extension of LOCRM12 (TITE‐LOCRM12) that accommodates both late‐onset toxicity and efficacy outcomes. The local modeling framework, which uses data from neighboring dose combinations, is robust and efficient when data are sparse in complex combination settings. Simulation studies show that the proposed design can achieve a high percentage of correctly selecting the OBDC while maintaining a feasible trial duration.
- 30 September 2026
- Published by Wiley in Pharmaceutical Statistics
- Vol. 25 (6) , e70120
TITE ‐ LOCRM12 : A Local Continual Reassessment Method for Drug Combination Optimization Based on Late‐Onset Toxicity and Efficacy Outcomes
The primary goal of Phase I/II combination trials is to identify the optimal biological dose combination (OBDC), defined as the dose combination that achieves the highest efficacy while maintaining an acceptable toxicity rate. Several designs have been proposed for such trials. Among them, the local continual reassessment method for Phase I/II trials (LOCRM12) performs well when compared with other model‐based methods in most of the scenarios studied. However, LOCRM12 assumes that both toxicity and efficacy outcomes are fully observed before enrolling the next cohort, an assumption that may not hold in real‐world settings, where late‐onset outcomes are increasingly common with the emergence of novel therapies such as molecularly targeted agents and immunotherapies. To address this limitation, we propose a time‐to‐event extension of LOCRM12 (TITE‐LOCRM12) that accommodates both late‐onset toxicity and efficacy outcomes. The local modeling framework, which uses data from neighboring dose combinations, is robust and efficient when data are sparse in complex combination settings. Simulation studies show that the proposed design can achieve a high percentage of correctly selecting the OBDC while maintaining a feasible trial duration.
- Open Access
- 1 March 2026
- Vol. 45 (6-7) , e70345
Applying the Partial Order Continual Reassessment Method to High-Dimensional Treatment Combinations.
Recent years have seen increasing popularities in therapies that combines several drugs and/or schedules, which imposes great difficulties on the design of appropriate Phase I clinical trials. The Partial Ordering Continual Reassessment Method (POCRM), among others, is a popular design that can be easily adapted to those complex settings. However, the design is introduced and evaluated in the setting of the combination of two drugs, and the combination of more than two drugs has rarely been investigated. In particular, the specification of toxicity orderings has always been an essential part of the POCRM, which becomes more difficult in high-dimensional settings due to the combinatorially increasing nature of the number of possible orderings of combination/schedules. This article proposes a systematic approach to specify orderings based on asymptotic properties in the setting of the combination of more than two drugs. Large simulation studies show that this novel ordering specification method leads to better design performance both asymptotically and with finite sample sizes.
- 1 March 2026
- Vol. 45 (6-7) , e70345
Open AccessApplying the Partial Order Continual Reassessment Method to High-Dimensional Treatment Combinations.
Recent years have seen increasing popularities in therapies that combines several drugs and/or schedules, which imposes great difficulties on the design of appropriate Phase I clinical trials. The Partial Ordering Continual Reassessment Method (POCRM), among others, is a popular design that can be easily adapted to those complex settings. However, the design is introduced and evaluated in the setting of the combination of two drugs, and the combination of more than two drugs has rarely been investigated. In particular, the specification of toxicity orderings has always been an essential part of the POCRM, which becomes more difficult in high-dimensional settings due to the combinatorially increasing nature of the number of possible orderings of combination/schedules. This article proposes a systematic approach to specify orderings based on asymptotic properties in the setting of the combination of more than two drugs. Large simulation studies show that this novel ordering specification method leads to better design performance both asymptotically and with finite sample sizes.
- Open Access
- 27 February 2026
- Published by Wiley in Statistics in Medicine
- Vol. 45 (6-7)
Applying the Partial Order Continual Reassessment Method to High‐Dimensional Treatment Combinations
Recent years have seen increasing popularities in therapies that combines several drugs and/or schedules, which imposes great difficulties on the design of appropriate Phase I clinical trials. The Partial Ordering Continual Reassessment Method (POCRM), among others, is a popular design that can be easily adapted to those complex settings. However, the design is introduced and evaluated in the setting of the combination of two drugs, and the combination of more than two drugs has rarely been investigated. In particular, the specification of toxicity orderings has always been an essential part of the POCRM, which becomes more difficult in high‐dimensional settings due to the combinatorially increasing nature of the number of possible orderings of combination/schedules. This article proposes a systematic approach to specify orderings based on asymptotic properties in the setting of the combination of more than two drugs. Large simulation studies show that this novel ordering specification method leads to better design performance both asymptotically and with finite sample sizes.
- 27 February 2026
- Published by Wiley in Statistics in Medicine
- Vol. 45 (6-7)
Open AccessApplying the Partial Order Continual Reassessment Method to High‐Dimensional Treatment Combinations
Recent years have seen increasing popularities in therapies that combines several drugs and/or schedules, which imposes great difficulties on the design of appropriate Phase I clinical trials. The Partial Ordering Continual Reassessment Method (POCRM), among others, is a popular design that can be easily adapted to those complex settings. However, the design is introduced and evaluated in the setting of the combination of two drugs, and the combination of more than two drugs has rarely been investigated. In particular, the specification of toxicity orderings has always been an essential part of the POCRM, which becomes more difficult in high‐dimensional settings due to the combinatorially increasing nature of the number of possible orderings of combination/schedules. This article proposes a systematic approach to specify orderings based on asymptotic properties in the setting of the combination of more than two drugs. Large simulation studies show that this novel ordering specification method leads to better design performance both asymptotically and with finite sample sizes.
- Open Access
- 23 December 2025
- Published by Wiley in European Journal of Haematology
- Vol. 116 (4) , 416-424
Improvements in Real‐World Survival in the Setting of a Recent Paradigm Shift in Acute Myeloid Leukemia Treatment
Therapeutic options for newly diagnosed (ND) acute myeloid leukemia (AML) have increased in recent years, leading to a shift in the treatment paradigm from conventional, intensive chemotherapy toward targeted and less intensive therapy. Since 2017, there has been a surge in FDA approvals for novel therapies, including small molecule inhibitors and new combinations of chemotherapy and targeted drugs. It is unclear what effect these changes have had on overall survival in the real‐world clinical setting over time. We retrospectively analyzed 501 patients with ND AML treated at an urban academic cancer center from 2015 to 2022 with respect to survival rates and treatment regimen utilization trends over time, in the context of AML risk category and whether allogeneic stem cell transplantation (HSCT) occurred. Among European LeukemiaNet 2022 intermediate risk ND AML patients who did not undergo HSCT, overall survival and relapse‐free survival improved significantly in the early versus later cohort (2015–2018 vs. 2019–2022, respectively). These gains were not observed in the adverse‐risk subgroup, where outcomes remain unchanged. First‐line use of hypomethylating agent‐based therapies increased seven‐fold, while reliance on chemotherapy alone decreased by half. These findings highlight a meaningful shift in treatment patterns associated with improved outcomes overall, but an ongoing lack of progress in the highest‐risk subgroups.
- 23 December 2025
- Published by Wiley in European Journal of Haematology
- Vol. 116 (4) , 416-424
Open AccessImprovements in Real‐World Survival in the Setting of a Recent Paradigm Shift in Acute Myeloid Leukemia Treatment
Therapeutic options for newly diagnosed (ND) acute myeloid leukemia (AML) have increased in recent years, leading to a shift in the treatment paradigm from conventional, intensive chemotherapy toward targeted and less intensive therapy. Since 2017, there has been a surge in FDA approvals for novel therapies, including small molecule inhibitors and new combinations of chemotherapy and targeted drugs. It is unclear what effect these changes have had on overall survival in the real‐world clinical setting over time. We retrospectively analyzed 501 patients with ND AML treated at an urban academic cancer center from 2015 to 2022 with respect to survival rates and treatment regimen utilization trends over time, in the context of AML risk category and whether allogeneic stem cell transplantation (HSCT) occurred. Among European LeukemiaNet 2022 intermediate risk ND AML patients who did not undergo HSCT, overall survival and relapse‐free survival improved significantly in the early versus later cohort (2015–2018 vs. 2019–2022, respectively). These gains were not observed in the adverse‐risk subgroup, where outcomes remain unchanged. First‐line use of hypomethylating agent‐based therapies increased seven‐fold, while reliance on chemotherapy alone decreased by half. These findings highlight a meaningful shift in treatment patterns associated with improved outcomes overall, but an ongoing lack of progress in the highest‐risk subgroups.