Personalized Health Care (PHC) in Cancer - Genomic Biomarkers for Pharmaceutical Development - page 29

TABLE 2.2 FdA Criteria for Retrospective Biomarker Analyses

The trial must be adequate, well-conducted and well-controlled.

Tumor tissue must be obtained in ≥ 95% of the registered and randomized study subjects and an evaluable result
(presence of wild-type or mutant KRAS) must be available for ≥ 90% of the registered and randomized study
subjects.

Before analysis, the FDA must have reviewed the assay methodology and determined that it has acceptable
analytical performance characteristics (for example, sensitivity, specificity, accuracy, precision) under the
proposed conditions for clinical use.

Before analysis of clinical outcomes based on the genetic testing, agreement with the FDA must be reached on the
analytical plan and promotional claims.

Sample size must be sufficiently large to be likely to ensure random allocation to each of the study arms for
factors that were not used as stratification variables for randomization.

Genetic analysis must be performed according to the qualified assay method by individuals who are masked to
treatment assignment and clinical outcome results.

predictive biomarker for poor response to EGFR mAbs [35,39-41] . Although post hoc evalu-
ation of clinical data is not generally accepted by the FDA, it acknowledged in this case that
incorporating this predictive marker into routine clinical practice would benefit patients by
sparing them treatment unlikely to be efficacious. Accordingly, the FDA has updated the
labels of EGFR mAbs to indicate that administration is not recommended in patients with
mutant KRAS. Additionally, the European Medicines Agency (EMEA) approved panitu-
mumab contingent on the requirement for determining patient KRAS status prior to admin-
istration [39,40,42] . This situation of retrospective analyses leading to additional biomarker
discovery is not unique to EGFR mAbs, and as such, the FDA has provided guidance
regarding the criteria that need to be met before it will consider the results of these types of
analyses ( Table 2.2 ). This is encouraging for other therapies that may be able to find similar
relationships and reflects the overall understanding that there can be potential value in ret-
rospective analyses, provided the conditions are appropriate.

EGFR TKI THERAPY IN NSCLC
As described for CRC, the EGFR is also expressed in NSCLC where it has been correlated
with aggressive disease and decreased survival [35] . Similar to what was seen in CRC with
EGFR mAbs, EGFR expression does not correlate with response to EGFR TKIs. In this case,
however, mutations in the target itself, EGFR, were shown to predict response to EGFR TKIs
in NSCLC [43-46] . The majority of these mutations were observed in two hotspots: in frame
deletions in exon 19 and a point mutation involving the replacement of leucine with argi-
nine at codon 858 (L858R) in exon 21 [33,44] . EGFR activating mutations result in ligand-
independent EGFR signaling, leading to uncontrolled tumor proliferation. EGFR gene copy
number has also been associated with response to TKIs, but with lower sensitivity and
specificity than mutations [33] . Together, clinical studies have shown that 55-82% of patients
with EGFR activating mutations responded to TKIs, compared with a 20-30% response rate
to platinum-based chemotherapy [46-50] . In contrast, those lacking EGFR activating muta-
tions rarely benefit from TKIs [46-50] . Specifically, in a trial by the West Japan Oncology