Kidney Cancer Journal 45
portant clinical implications by helping to select patients
that will benefit from a more proactive management.
A vast number of histopathologic, clinical and biochemical
parameters have been identified to be prognostic.
These include: bone metastasis, liver metastasis, elevated
neutrophil-to-lymphocyte ratio, elevated C-reactive protein,
nonclear cell RCC histology, papillary RCC histology,
high body mass index, brain metastasis and renal
dysfunction.19 However, at this point, none of them have
been shown to improve the prognostic performance of
the traditional models and the best way to incorporate
them in clinical practice remains to be determined.
Predictive Capability of IMDC
Prognostic Model: CheckMate 214
As the experience with targeted therapies and IO therapy
unfolds, the IMDC model proves to have, not only a prognostic
ability, but also a predictive capability. The predictive
capability was demonstrated in the CheckMate
214 trial, a phase 3 study evaluating the combination of
nivolumab and ipilimumab versus standard VEGF-targeted
therapy with sunitinib alone in treatment-naïve
mRCC.1 This pivotal report stratified patients by IMDC
prognostic group. The highlights of this trial are that in
the IMDC intermediate- and poor-risk groups, combination
IO therapy was associated with superior OS as compared
with sunitinib (HR, 0.66, P<0.0001). PFS was also
improved in these patients (HR, 0.82, P=0.03), but the result
was not statistically significant per the pre-specified
threshold (P=0.009). On the contrary, in the IMDC favorable
risk group, superior outcomes were observed with sunitinib
alone in terms of objective response rate (29% vs
52%, P<0.001), PFS (HR1.23, P=1.888) and OS (HR 1.22,
P=0.4426).20
These provocative results raise at least two intriguing
issues. First, the fact that the intermediate/poor risk
groups patients and the favorable group patients derived
different benefits from the nivolumab/ipilimumab combination
and sunitinib might suggest that their tumor
biology is different. A plausible hypothesis is that the favorable
risk group tumor has a greater addiction to the
VEGF pathway. In other tumor sites such as EGFR-mutated
non-small cell lung cancer, ICP was also demonstrated
to be less effective than TTx.21 This leads us to believe that
in tumors with a strong angiogenic dependency, using
ICP alone is less effective and insufficient to halt tumor
growth. Another possible explanation might be related to
the immunogenicity of the tumor and its microenvironment.
This important question requires to be further elucidated.
Recently, the results of Javelin Renal 101 and
Keynote-426, two pivotal phase III trials that combine
ICP/VEGF-targeted therapy, demonstrated that the combination
of ICP and TTx also benefits the favorable-risk
group RCC suggesting that the addition of targeted therapy
to ICP can prime immune response.2,3 Furthermore,
the other issue raised by CHECKMATE-214, although not
specific to RCC, is that PFS and OS are not strongly correlated
in trials using ICP alone.1 The surrogacy of PFS for
OS is therefore weak with this class of treatment. The reason
for this remains unclear and might imply that the antitumor
effect of ICP likely lasts well beyond its admin-
istration duration, extending to subsequent therapies,
and acting in synergy. In the particular context of immunotherapy,
possible new prognostic and predictive factors
might emerge over the next year to improve the traditional
model.
Integrating Molecular and Genomic
Factors in Prognostication
Advances in sequencing technologies have opened up
new avenues of investigation, a development that is providing
exciting insights into the molecular landscape of
mRCC. Equally exciting is the potential of integrating
genomic or molecular markers to the IMDC model and
other prognostic tools to guide better treatment strategies.
Large scale sequencing analysis has revealed that the
most frequent genomic abnormalities found in predominantly
early RCC patients are: the inactivation of the VHL
tumor suppressor gene, alterations of the 3p chromatin
modulators/modifiers (eg, PBRM1, SETD2, BAP1, KDM5C,
KMD6A), alterations of the p53 signaling and alterations
of the PI3K/AKT/mTOR pathway, recurrent arm level or
focal losses on chromosomes 1p, 3p, 4q, 6q, 8p, 9p, and
14q, and gains on chromosomes 1q, 2q, 5q, 7q, 8q, 12p,
and 20q.22,23
Voss et al. explored the correlation between frequent
mutations found in early RCC and OS in patients with
mRCC.24 Using a multivariate Cox regression analysis,
BAP1, TP53 and PBRM1 mutations were demonstrated to
be independently associated with OS in patients enrolled
in the phase III COMPARZ (training set). Mutation status
was added to the MSKCC model in an effort to enhance
its performance. The analysis showed that the MSKCC
model’s c-index was improved by the addition of the genomic
information in the training set (from 0.595 to
0.628) as well as the validation set (from 0.622 to 0.641),
composed of patients from the phase II RECORD3 trial.
The clinical relevance of this difference is unclear and
more studies are required to establish the predictive utility
of adding genomic information.
Similarly, in a smaller dataset, Bosse et al. performed a
study using genomic data to refine the IMDC model.25
The association between OS and alterations in PBRM1,
BAP1, SETD2, KDM5C and TP53 was evaluated. Only
BAP1 or BAP1 and TP53 combined correlated with worse
OS, in multivariate models. Stratifying by IMDC risk
groups, only patients in the IMDC poor risk group have
a statistically significant worse survival while carrying
mutations in BAP1 or BAP1 and TP53 combined.
Overall, this evidence supports the idea that genomic
information, originally discovered in early RCC, might
have a prognostic significance for mRCC patients and
that its integration in traditional prognostic models
might be suitable. However, external validation in larger
datasets is still warranted. Furthermore, in these two abstracts,
another important caveat to consider is that it is
unclear if the mutational status was assessed on the primary
tumor or metastasis. Taking into account intratumoral
heterogeneity and tumor evolution, one might
consider the analysis of metastasis tissue more suitable in
this setting.
Gene-expression signatures for prognosis may add
complementary prognostic information in the metastatic
setting. In this line of thinking, De Velasco et al. tested