An exploratory scoring system (“Matching Score”) was
developed, as previously described.
6 (link),9 (link) The Matching
Score was calculated post hoc by investigators blinded to outcomes at the time
and it was based upon the actual drugs administered. Under this system, the
higher the Matching Score, the better the match. In general, the Matching Score
was calculated by dividing the number of alterations matched in each patient
(numerator) by the number of characterized aberrations in that patient’s
tumor (denominator). For instance, if a patient’s tumor harboring six
genomic aberrations received two drugs that targeted three of the
patient’s genomic alterations, the Matching Score would be 3/6 or 50%.
This is because certain drugs targeted more than one alteration (e.g., many
small molecule inhibitors often have activity against multiple kinases) and were
counted as matches for each identified genomic alteration that was matched.
Other considerations were as follows:
two mutations in the same gene that had the same effect
(e.g., loss of function) counted as one aberration in the
denominator; two mutations in the same gene that were known to
function differently counted twice.
two different structural alterations in the same gene (e.g.,
amplification and mutation) were counted as two aberrations in the
denominator since they have different functional effects (e.g.,
overexpression versus activation);
two drugs targeting the same alteration were counted twice
in both the numerator and denominator if they had well-established
synergy (e.g. the FDA-approved combinations of dabrafenib and
trametinib for BRAF mutations, or pertuzumab and
trastuzumab ERBB2 alterations);
only if the patient was matched (in part) based on hormone
(ER) positivity in the tissue biopsied for genomic analysis, the ER
status was then added to both the numerator and the denominator;
all variants of unknown significance were excluded;
in the case of cell cycle inhibitors that targeted CDK4/6,
we counted any concomitant CDK4/6 and
CDKN2A/B alterations (N=2 patients) or
CCND1/2/3 and CDKN2A/Balterations (N=2 patients) as one alteration and one drug target in
the numerator and denominator, because the CDKN2A protein,
p16(INK4a), directly binds to the CDK4/CDK6/Cyclin D1 complex, thus
regulating their activity.39 (link),40 (link)
TP53 alterations were considered matched to
anti-angiogenic agents, based on data showing that
TP53 mutations are associated with upregulation
of VEGF-A and that treatment of TP53-mutant tumors
with anti-angiogenic agents is associated with improved
outcomes.27 (link),28 ,41 (link),42 (link)
if the patient was treated with immunotherapy (e.g.,
anti-PD-1 or anti-PD-L1 checkpoint inhibitors), the Matching Score
was 100% for PD-L1 IHC high positive, TMB high, MSI high results (or
MHL1, MSH2,
MSH6, PMS2 alterations), or if
none of the aforementioned were known, but the patient had ≥8
genomic alterations (N=1 patient) based upon the assumption of a
high TMB.
if PD-L1 IHC was low positive, the TMB was intermediate, or
there was a CD274 (PD-L1) amplification, the
Matching Score was 50%; if the patient received a combination of a
checkpoint inhibitor and a gene-targeted drug that matched one or
more of his/her genomic alterations, the score was >50%. As
an example, if a patient had intermediate TMB and a
MET amplification, as well as a
TP53 mutation, and was treated with nivolumab
and the MET inhibitor, crizotinib, the Matching Score would be
>50%.
if more than one NGS report was available, the alterations
in each report were counted (since there can be heterogeneity
between tissue biopsies);
if a patient’s regimen included drugs that did not
match any alteration, those drugs received a Matching Score of
0.
The cut-off of 50% for the analyses of low versus high Matching Scores
was chosen according to the minimum P-value criteria.
19 (link) See
Supplemental Text for selected
examples of therapy and Matching Score methodology.
Sicklick J.K., Kato S., Okamura R., Schwaederle M., Hahn M.E., Williams C.B., De P., Krie A., Piccioni D.E., Miller V.A., Ross J.S., Benson A., Webster J., Stephens P.J., Lee J.J., Fanta P.T., Lippman S.M., Leyland-Jones B, & Kurzrock R. (2019). Molecular profiling of cancer patients enables personalized combination therapy: the I-PREDICT study. Nature medicine, 25(5), 744-750.