Molecular Profiling benefits not proven in CUP study but…

January 28, 2020

From March 2012 to February 2018, 243 CUP patients from 4 EU countries took part in a Phase III trial (GEFCAPI 04) led by Dr Fizazi in France. The aim was to see if tailored treatment following molecular profiling would improve outcomes; but the results, reported in the Autumn of 2019, showed no significant improvement in outcomes overall. There was , however, a substantial survival advantage for a minority of patients with more treatable cancer types.

Taken superficially the overall outcome could seem to be a major blow to the supporters of the potential benefits of molecular profiling; but the headline result should be viewed with caution. World-leading CUP authority, Dr Tony Greco, comments (Jan 2020): ‘I am more optimistic about molecular testing in CUP than ever, particularly since so many previously nonresponsive/poorly treated advanced cancers are becoming more and more treatable/responsive.’

We asked Dr Greco for his specific comments on the trial – shown below – and his analysis illustrates how the headline outcome needs to be set against the methodology and the detailed findings:

This phase 3 randomized study thus far published as an abstract randomized patients with CUP to receive either site specific determined by the molecular assay diagnosis versus empiric chemotherapy with gemcitabine/cisplatin.. The majority were tested with CancerTYPE ID-222 of 243 patients and accrued from 03/12 to 02/18. The study was designed in 2010-2011 and site specific therapies reflect the standard of 2010-2011. It is important to presume, which I think is without a doubt, that CUP represents many specific cancer types with a biology similar to those cancer types rather than a single entity or cancer type, although all have no anatomical primary site identified (most are found at autopsy). The only biologic difference is the size of the primary and “early” metastasis.  I will list various issues:

1. The majority of the patients (about 75%) with CUP entered into this trial had molecular diagnoses of relatively resistant cancer types and the most common included pancreaticobiliary (includes cholangiocarcinoma, pancreas, gall bladder), advanced metastatic squamous cell carcinoma (several sites), GE junction and gastric carcinoma and a few others. Why so many relatively resistant cancers? I think the answer is patient selection (see 4. below). Site specific treatment for most of these cancer types during the time of this study (and even now for some of these cancers) was poor and would not be expected to be any better than empiric chemotherapy with gemcitabine/cisplatin.
2. The minority (60 patients) had molecular diagnoses of more responsive cancer types (renal, colorectal, non-small cell lung, breast, melanoma, ovarian, bladder, others) that would be expected to benefit more from site specific therapy than empiric chemotherapy.  In this group of patients in this study 37 were randomly assigned to site specific treatment and 23 received empiric chemotherapy. This small number precludes adequate statistical comparison of median survivals (under powered). However at 1 year 39% of the site specific group were alive versus 30% in the empiric group and at 2 years 24% alive versus 10%.  The power of this observation is also poor since there were only 60 patients.  Other small prospective studies as well as larger retrospective and phase 2 prospective nonrandomized studies show the same results in this group of patients with molecular diagnoses of sensitive cancer types, favoring site specific therapy.
3. The overall results of this study did not show a median survival advantage for site specific treatment for the all the patients is entered in this study.
4. The population of patients diagnosed with CUP in clinical practice has changed substantially in the last 10 years due to improved immunohistochemical (IHC) based diagnostics.  This has restricted the definition of CUP by many oncologists and some of the more sensitive cancer types diagnosed by IHC (lung, breast, colorectal, renal, others)are often treated as such and not considered/entered on clinical trials. Other cancer types for which specific IHC protein markers are not available for precise diagnosis (such as pancreaticobiliary carcinomas, squamous cell carcinomas, other upper GI carcinomas and others are more likely to be entered into clinical trials such as GEFCAP1-04. This happened.  In essence many of the more treatable cancers were not included in this study since the oncologists seeing these patients with more specific immunohistochemical markers indicating a specific more treatable/responsive  cancer type were not entered in the study but treated off study with site specific therapies.
5. In this study the site specific therapy for each molecular diagnosis was determined in 2010-2011 and was not changed during the study period 2012-2018.  There were many patients who could not benefit from important improvements in treatment developed in the last decade including various targeted therapy and immunotherapy.  Certain patient subgroups did not receive the standard newer therapies but rather received inferior standard of care from 2010-2011. There has been rapid change/improvements with immune checkpoint inhibitors and novel targeted therapies for kidney cancer, melanoma, non-small cell lung cancer, colorectal, urothelial carcinoma, GE junction/gastric carcinomas, breast cancer, and others. Patients with these molecular diagnoses did not receive the newer more effective site specific therapies in this study.
6. For patients who are molecularly diagnosed to have relatively unresponsive/insensitive cancer types as defined today, it is reasonable to use either empiric chemotherapy or site specific therapy but both produce relatively poor results. There is no way to determine in CUP patients ( with the exception of the minority diagnosed by IHC) if they have sensitive or insensitive cancer types without a molecular classifier assay.
7. Future randomized controlled trials of standard contemporary therapy for patients with several advanced cancers, particularly the more responsive cancer types, are needed to accurately address the value of a specific diagnosis of the cancer type in CUP patients.
8. In summary CUP is not a single metastatic disease but an extremely heterogeneous group of patients representing many advanced cancers and subtypes.  Some of these patients have highly treatable cancer and others do not.  As time goes on more of the unresponsive/relatively untreatable cancer types become much more treatable. This happened rather rapidly for many advanced cancer types in the last 10 years and the expectations are that recognition and appropriate precision therapy for other patients will continue and also improve their prognosis.  The GEFCAP1-4 study suffered from many flaws as a result of patient selection, patient heterogeneity and the evolving improved therapy found in the last decade for several subgroups .  Nonetheless, in the minority of patients diagnosed with more treatable cancer types there was a substantial survival advantage at 2 years for site specific therapy  based on the molecular assay versus empiric chemotherapy, albeit this was a relatively small group.

← Back to blog