Limits...
Next-generation clinical trials: Novel strategies to address the challenge of tumor molecular heterogeneity.

Catenacci DV - Mol Oncol (2014)

Bottom Line: The promise of 'personalized cancer care' with therapies toward specific molecular aberrations has potential to improve outcomes.These issues have become hurdles to advancing cancer treatment outcomes with novel molecularly targeted agents.Finally intra-patient heterogeneity through time may be addressed by serial biomarker assessment at the time of tumor progression.

View Article: PubMed Central - PubMed

Affiliation: University of Chicago Medical Center, Department of Medicine, Section of Hematology & Oncology, 5841 S. Maryland Avenue, MC2115, Chicago, IL 60637, USA. Electronic address: dcatenac@medicine.bsd.uchicago.edu.

Show MeSH

Related in: MedlinePlus

Inter–patient tumor molecular heterogeneity. (A) Genomic profiling using a ~240 gene next-generation sequencing (NGS) platform of a cohort of 50 stage IV GEC samples (upper panel) revealing few high frequency events (peak) and numerous low frequency events (tail); pie chart revealing profound inter-patient molecular heterogeneity (see Table 3). (Catenacci et al., 2014a) (B) Proteomic expression profiling of 100 GEC samples using multi-plex (8 peptides shown) selected reaction monitoring (SRM) mass spectrometry (MS) revealing clear inter-patient heterogeneity. (Catenacci et al., 2014a,b; Hembrough et al., 2012).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4402102&req=5

Figure 3: Inter–patient tumor molecular heterogeneity. (A) Genomic profiling using a ~240 gene next-generation sequencing (NGS) platform of a cohort of 50 stage IV GEC samples (upper panel) revealing few high frequency events (peak) and numerous low frequency events (tail); pie chart revealing profound inter-patient molecular heterogeneity (see Table 3). (Catenacci et al., 2014a) (B) Proteomic expression profiling of 100 GEC samples using multi-plex (8 peptides shown) selected reaction monitoring (SRM) mass spectrometry (MS) revealing clear inter-patient heterogeneity. (Catenacci et al., 2014a,b; Hembrough et al., 2012).

Mentions: The ToGA trial evaluated trastuzumab for ‘HER2 positive’ GEC, (Bang et al., 2010) and screened 3803 patients internationally to obtain 810 eligible patients, of which 594 were otherwise eligible for randomization (Table 2). To be eligible, ‘HER2 positive’ was defined as a ‘FISH+’ ratio ≥2 with any IHC score (0–3+), or IHC3+ with ‘FISH-‘; ‘IHC2+/FISH-‘ patients were ineligible. After excluding ineligible patients by these biomarker screening criteria, as well as those not meeting other trial entry criteria, only 15.6% of all-comers with stage IV GEC were eligible for therapy. Based on preplanned subset analyses, ‘HER2 positivity’ is now clinically defined with a more stringent threshold than even ToGA initially used for screening: (IHC2+/FISH+, IHC3+/anyFISH), which would exclude the 131 patient tumors with FISH+/IHC0-1+ scores who appeared to derive no benefit from the addition of trastuzumab. That leaves 463 patients from the original 3803 screened patients (12%), or 57% of the initially identified ‘HER2+ patients’ in the trial. By acknowledging the disappointments of applying targeted therapies in a ‘one-size-fits-all’ strategy, the ToGA trial illustrates the ongoing challenge when attempting to select patients for targeted therapies. This includes the extremely high numbers of patients required to screen when attempting to apply classic clinical trial designs, with frequentist statistical methods, (Simon and Maitournam, 2004) to subsets within a very molecularly heterogeneous disease such as GEC. Worse, the example of HER2, entailing ~10–15% of GEC, is one of the larger ‘slices of the pie’ (Figure 3, Table 3). The accrual numbers that were required for the ToGA trial demonstrates how profound inter-patient molecular heterogeneity is challenging the application of novel targeted agents for specific sub-populations using traditional clinical trial designs. Selecting patients with MET amplified tumors at ~4% incidence within GEC for anti-MET therapy, (Smolen et al., 2006; Catenacci et al., 2011b, 2014b) which is based on sound preclinical and clinical evidence, is an even more difficult challenge than the HER2 ToGA example. Such a phase III trial would require >15000 total GEC patients with stage IV disease to be screened to accomplish a ‘MET amplified‘ phase III selection trial. When also considering that there are several redundant drugs adopting the same strategy for this limited patient cohort, a large randomized phase III trial is seemingly impossible (as is even a randomized phase IIb). Importantly, there is increasing recognition of multiple rare molecular subsets (including both genomic events or proteomic expression ‘cut-offs’ that are considered biologically vital) within solid tumors. Therefore, selection strategies within biomarker-driven trials using ‘à la carte’ low throughput companion diagnostic assays, such as IHC, PCR or FISH, result in sizeable screening delays. (Stricker et al., 2011) With each diagnostic test having its own central site, serial eligibility screening for each individual trial is required – all while the patient awaits without therapy. Patient drop-out due to wait-time is high in this setting, unfortunately. Moreover, in the stage IV setting, the tumor sample is most often from a small tissue biopsy that is ultimately exhausted via repeated serial analyses, precluding further screening for trial eligibility without a repeat tissue biopsy. Further, the odds of qualifying for a given trial are low given the relative infrequency of each aberration. Crucially, multiple concurrent events (genomic and proteomic) within a given tumor biopsy, often not fully appreciated when presented as pie charts or ‘high-peak long-tail charts’, further complicate treatment stratification (Table 3). (Catenacci et al., 2014a) For instance, focusing solely on genomic aberrations without considering proteomic profiling, one tumor sample can range from having 0–18 identified ‘actionable’ events on medium-throughput NGS platforms (~250–300 genes). (Catenacci et al., 2014a; Sehdev and Catenacci, 2013b; Frampton et al., 2013) Which of the multiple genomic events should be targeted? How to navigate the infinite possible drug-combinations without established phase I data? What if the drug is not yet available commercially and no trial immediately available? How do we test the hypothesis of each actionable molecular event and ‘matched’ drug with statistical power to rule out effect from random variation (which to the extreme, ultimately approaches an ‘N-of-1’ trial)? (Parmigiani et al., 2009) The number of patients required, the limited amount of tissue available, and the length of time to results acquisition, along with the dilemma of multiple ‘actionable’ events in a given sample, all highlight serious challenges currently imposed by inter-patient tumor molecular heterogeneity.


Next-generation clinical trials: Novel strategies to address the challenge of tumor molecular heterogeneity.

Catenacci DV - Mol Oncol (2014)

Inter–patient tumor molecular heterogeneity. (A) Genomic profiling using a ~240 gene next-generation sequencing (NGS) platform of a cohort of 50 stage IV GEC samples (upper panel) revealing few high frequency events (peak) and numerous low frequency events (tail); pie chart revealing profound inter-patient molecular heterogeneity (see Table 3). (Catenacci et al., 2014a) (B) Proteomic expression profiling of 100 GEC samples using multi-plex (8 peptides shown) selected reaction monitoring (SRM) mass spectrometry (MS) revealing clear inter-patient heterogeneity. (Catenacci et al., 2014a,b; Hembrough et al., 2012).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4402102&req=5

Figure 3: Inter–patient tumor molecular heterogeneity. (A) Genomic profiling using a ~240 gene next-generation sequencing (NGS) platform of a cohort of 50 stage IV GEC samples (upper panel) revealing few high frequency events (peak) and numerous low frequency events (tail); pie chart revealing profound inter-patient molecular heterogeneity (see Table 3). (Catenacci et al., 2014a) (B) Proteomic expression profiling of 100 GEC samples using multi-plex (8 peptides shown) selected reaction monitoring (SRM) mass spectrometry (MS) revealing clear inter-patient heterogeneity. (Catenacci et al., 2014a,b; Hembrough et al., 2012).
Mentions: The ToGA trial evaluated trastuzumab for ‘HER2 positive’ GEC, (Bang et al., 2010) and screened 3803 patients internationally to obtain 810 eligible patients, of which 594 were otherwise eligible for randomization (Table 2). To be eligible, ‘HER2 positive’ was defined as a ‘FISH+’ ratio ≥2 with any IHC score (0–3+), or IHC3+ with ‘FISH-‘; ‘IHC2+/FISH-‘ patients were ineligible. After excluding ineligible patients by these biomarker screening criteria, as well as those not meeting other trial entry criteria, only 15.6% of all-comers with stage IV GEC were eligible for therapy. Based on preplanned subset analyses, ‘HER2 positivity’ is now clinically defined with a more stringent threshold than even ToGA initially used for screening: (IHC2+/FISH+, IHC3+/anyFISH), which would exclude the 131 patient tumors with FISH+/IHC0-1+ scores who appeared to derive no benefit from the addition of trastuzumab. That leaves 463 patients from the original 3803 screened patients (12%), or 57% of the initially identified ‘HER2+ patients’ in the trial. By acknowledging the disappointments of applying targeted therapies in a ‘one-size-fits-all’ strategy, the ToGA trial illustrates the ongoing challenge when attempting to select patients for targeted therapies. This includes the extremely high numbers of patients required to screen when attempting to apply classic clinical trial designs, with frequentist statistical methods, (Simon and Maitournam, 2004) to subsets within a very molecularly heterogeneous disease such as GEC. Worse, the example of HER2, entailing ~10–15% of GEC, is one of the larger ‘slices of the pie’ (Figure 3, Table 3). The accrual numbers that were required for the ToGA trial demonstrates how profound inter-patient molecular heterogeneity is challenging the application of novel targeted agents for specific sub-populations using traditional clinical trial designs. Selecting patients with MET amplified tumors at ~4% incidence within GEC for anti-MET therapy, (Smolen et al., 2006; Catenacci et al., 2011b, 2014b) which is based on sound preclinical and clinical evidence, is an even more difficult challenge than the HER2 ToGA example. Such a phase III trial would require >15000 total GEC patients with stage IV disease to be screened to accomplish a ‘MET amplified‘ phase III selection trial. When also considering that there are several redundant drugs adopting the same strategy for this limited patient cohort, a large randomized phase III trial is seemingly impossible (as is even a randomized phase IIb). Importantly, there is increasing recognition of multiple rare molecular subsets (including both genomic events or proteomic expression ‘cut-offs’ that are considered biologically vital) within solid tumors. Therefore, selection strategies within biomarker-driven trials using ‘à la carte’ low throughput companion diagnostic assays, such as IHC, PCR or FISH, result in sizeable screening delays. (Stricker et al., 2011) With each diagnostic test having its own central site, serial eligibility screening for each individual trial is required – all while the patient awaits without therapy. Patient drop-out due to wait-time is high in this setting, unfortunately. Moreover, in the stage IV setting, the tumor sample is most often from a small tissue biopsy that is ultimately exhausted via repeated serial analyses, precluding further screening for trial eligibility without a repeat tissue biopsy. Further, the odds of qualifying for a given trial are low given the relative infrequency of each aberration. Crucially, multiple concurrent events (genomic and proteomic) within a given tumor biopsy, often not fully appreciated when presented as pie charts or ‘high-peak long-tail charts’, further complicate treatment stratification (Table 3). (Catenacci et al., 2014a) For instance, focusing solely on genomic aberrations without considering proteomic profiling, one tumor sample can range from having 0–18 identified ‘actionable’ events on medium-throughput NGS platforms (~250–300 genes). (Catenacci et al., 2014a; Sehdev and Catenacci, 2013b; Frampton et al., 2013) Which of the multiple genomic events should be targeted? How to navigate the infinite possible drug-combinations without established phase I data? What if the drug is not yet available commercially and no trial immediately available? How do we test the hypothesis of each actionable molecular event and ‘matched’ drug with statistical power to rule out effect from random variation (which to the extreme, ultimately approaches an ‘N-of-1’ trial)? (Parmigiani et al., 2009) The number of patients required, the limited amount of tissue available, and the length of time to results acquisition, along with the dilemma of multiple ‘actionable’ events in a given sample, all highlight serious challenges currently imposed by inter-patient tumor molecular heterogeneity.

Bottom Line: The promise of 'personalized cancer care' with therapies toward specific molecular aberrations has potential to improve outcomes.These issues have become hurdles to advancing cancer treatment outcomes with novel molecularly targeted agents.Finally intra-patient heterogeneity through time may be addressed by serial biomarker assessment at the time of tumor progression.

View Article: PubMed Central - PubMed

Affiliation: University of Chicago Medical Center, Department of Medicine, Section of Hematology & Oncology, 5841 S. Maryland Avenue, MC2115, Chicago, IL 60637, USA. Electronic address: dcatenac@medicine.bsd.uchicago.edu.

Show MeSH
Related in: MedlinePlus