Presented this morning at the American Association of Cancer Research meetings in Denver, Colorado, is an interesting study by Daniel Von Hoff and associates of the Phoenix-based Translational Genomics Research Institute (TGen).
From the TGen press release:
66 patients were treated at 9 different U.S. medical centers. All of the patients had previously experienced growth of their tumors while undergoing as many as two to six prior cancer treatments, including conventional chemotherapy.
However, after molecular profiling identified precise targets, new treatments were administered that resulted in patients experiencing significant periods of time when there was no progression of their cancer.
“This clinical trial was unique because patients acted as their own control,” said Dr. Von Hoff. “We compared each patient’s progression-free survival, following treatment based on molecular profiling, to how their tumors progressed under their prior treatment regimens, before molecular profiling.”
In a significant number of patients, the targeted treatments provided significantly longer periods when tumors did not progress, or even shrunk, said Dr. Von Hoff, who also is a Medical Director of US Oncology and a former Director of the Arizona Cancer Center at the University of Arizona.
He said this clinical trial demonstrated the value of personalized medicine, in which treatments are prescribed based on an individual’s specific genetic makeup. The type of drugs, dosages, their delivery and other treatment aspects – all are based on each patient’s individual medical needs.
Among the patients, 27 percent had breast cancer, 17 percent had colorectal cancer, 8 percent had ovarian cancer and 48 percent had cancers that were classified as miscellaneous.
Patients experienced varying levels of improvement. Among those with breast cancer, the period of progression-free survival increased for 44 percent of patients; for colorectal, 36 percent of patients; for ovarian, 20 percent of patients; and for miscellaneous cancers the improvement was seen in 16 percent of patients.
The molecular profiling for this research study was performed by Caris Diagnostics (Caris Dx) in Phoenix.
These results are the first in a series of studies in support of Target Now™, a commercially-available oncology testing service offered exclusively by Caris Dx. Target Now uses cutting-edge molecular profiling techniques, including both DNA microarray and immunohistochemical (IHC) analysis, to provide individualized information about a patient’s tumor as an aid to the treating oncologist.
The study was supported by a $5 million donation by a private philanthropist, apparently a cancer patient, as reported in the above press release.
Larry’s comment:
This is a very interesting and innovative study design. It is very difficult (and horrendously expensive) to perform prospective, randomized trials. It is much simpler to use each patient as his/her own control, comparing results (progression-free survival) of the previous, empiric therapy with the “targeted” therapy. Even with the simplified design, however, this pilot study cost $5 million. The weaknesses are the often poor correlation of progression-free survival with overall survival and the subjectivity of “progression free survival” as a study endpoint.
Study details are eagerly awaited.
Edit April 20, 2009 16:13 GMT –>
The Von Hoff study abstract is now on American Association for Cancer Research Website.
Larry’s comments, based on data in study abstract:
Patients eligible had to have (1) clear disease progression on prior therapy and (2) received at least two prior lines of systemic chemotherapy. “Success” was defined as a time to progression (TTP) of at least 30% longer than the TTP of the most immediate prior therapy. A total of 18 patients (of 66) achieved this 30% improvement in time to progression.
An interesting question is what percentage of cancer patients in general have increased times to progression with subsequent cycles of empiric chemotherapy, compared to the prior empiric therapy. As I write this, I don’t know if there are any published data on this. If anyone is aware of whether or not these data exist, I’d appreciate knowing about it. Either respond to this blog post (requires easy registration) or send me a private email: mail@weisenthal.org. Thanks!
An additional issue is the precision with which time to progression (TTP) with prior therapy was measured. With prospective clinical trials, using TTP as the primary endpoint, there is generally a protocol to standardize follow-up, with physical examination, laboratory tests, and radiographic tests being performed as standardized intervals, so that comparisons of TTP are meaningful. There would appear to be a problem with utilizing “historical” TTP on patients enrolled from the general oncology community, in which there is no standardization regarding the methodology for determining disease progression or the intervals at which these measurements are taking place. Additionally, we don’t know, from what is presented in the abstract, whether or not there was “blinded” assessment of TTP by outside auditors, or whether “historical” TTP was assessed and reported by the referring oncologists who enrolled the patients onto the trial.
Tags: Add new tag, American Association of Cancer Research, clinical trial, Daniel Von Hoff, DNA microarrays, Immunohistochemistry, Private philanthropy, progression-free survival, targeted therapy
The Von Hoff study is interesting not so much from the data it actually presents but rather that it indicates the interest for individualized cancer therapy. From the abstract it is not possible to know what has actually been done (what molecular profiling? how was treatment selected, what treatment was given, how was PFS assessed before/after the selected therapy, principles for quality assurance etc). The effect also seems modest.
However, using the patient as his/her own control using PFS as endpoint is a useful study design that could be used to assess predictive tests for drug selection when it is not possible to do a randomized trial or to provide a basis to start one. PFS inversion, ie longer PFS on treatment selected by test than the standard empirical treatment given just before would be a justified endpoint provided that PFS is accurately measured.
I would guess that some labs doing drug resistance test in vitro now has sufficiently long experience and reasonably a large number of patient cases (considerably more than Von Hoff’s 66) that has been treated with assay based therapy after progression on empirical standard. Why don’ t such a lab, eg Larry Weisenthal’s, put their data together along a defined protocol to keep quality and avoid bias, to show the effect from test selected therapy on ‘PFS inversion’? There is probably a difficulty in getting report backs from the clinicians but still this approach seems to be the easiest way forward at present to assess the utility of the cell based tests.
Larry Weisenthal’s proposal to show the effect of test selected therapy, ie a prospective trial in a number of tumor types at very high costs would, of course, be the ultimate proof of efficacy but will probably never be done. When it comes to how to proof the utility of predictive tests, the cell based in vitro tests might have been better off now if the ambition in earlier clinical studies had been more limited, eg focusing on one key drug in one tumor type, ie something similar to the K-ras mutation test for EGFR antibodies in colorectal cancer and that is now regarded as a break-through in the field of individualization despite its modest clinical importance. Trying to target all drugs in all major cancer types at the same time could have been a later approach to be developed gradually once the concept of in vitro drug tests had been accepted in one clinical situation.
- Peter Nygren (Uppsala, Sweden)
I heard from Robert Nagourney that he attended Von Hoff’s presentation at the American Association of Cancer Research meetings. Robert indicated that he’d provide some more details and his own observations, upon his return from Colorado. Von Hoff’s paper was presented at the Plenary Session, entitled “Spotlight on Breakthroughs in Cancer Research.” So this was considered to be of major importance, despite what would appear to be extremely modest results, as noted above by Peter Nygren.
My impression, based only on the abstract, is to be underwhelmed. They got a 30% increase or greater in time to progression in only 18 of 66 patients, based upon comparison to prior therapy. Firstly, it’s hard to believe that this would even be significant in a randomized trial vs empiric therapy, as I’m sure that some patients on third or 4th line empiric therapy have improved times to progression on subsequent 4th and 5th line empiric therapy. Second, as I wrote, TTP is a very soft and subjective endpoint, especially in the absence of rigid follow up protocols to allow for apples to apples comparisons with the control (previous or contemporaneous) therapy.
Peter is astute, however, in his observation that the Von Hoff trial design does set a precedent. The pharmaceutical industry has exploited “time to progression” as an outcomes endpoint with great success (to the goals of the pharmaceutical industry). This does open up a whole new avenue for obtaining elusive data on “efficacy.”
My own “practice” is fairly unique. For the past 17 years, I’ve been performing cell culture-based ITRT for a handful of referring oncologists, 4 of which have each sent me more than 500 referrals over the years (i.e. 2,000 cases from 4 oncologists). There are certainly many cases which would fulfill the Von Hoff entry criteria (disease progression after 2 prior forms of empiric chemotherapy). The problem is that these oncologists all tend to treat patients on the basis of ITRT results. So prior, empiric, therapy would have been administered by different oncologists than the oncologists who gave ITRT-directed therapy. So I have two problems: firstly, tracking down the prior oncologists and getting them to agree to cooperate in the absence of financial inducement. Secondly, the likelihood that the accuracy of “time to progression” with the prior, empiric therapy will not be consistently reliable (this is one of my criticisms of the Von Hoff study, offered above). But it’s certainly worth exploring — both by me and by others in the ITRT field.
Peter’s second suggestion is truly excellent, and it gives me an idea. He uses the KRAS example, stating that a focused effort to “validate” an ITRT assay for only a single class of drugs could be much more successful than historical efforts to “validate” ITRT for all drugs and all tumors.
The idea I have is for a collaborative paper, focusing on the platinums — arguably the most important drugs in all of cancer chemotherapy.
Here is what I know from the literature and from my own experience:
1. Many laboratories have reported good clinical correlations between assay results and response to chemotherapy in ovarian cancer, where virtually all chemotherapy is platinum-based.
2. Many laboratories have reported good clinical correlation between assay results and patients survival (progression-free survival and/or overall survival) in ovarian cancer.
3. A number of laboratories have reported that specimens from previously-untreated patients are more platinum-sensitive in the assays than are specimens from previously-treated patients.
4. Our own data show clear differences in platinum sensitivity in ovarian cancer patients meeting the definition of having “platinum-sensitive” disease (relapse free interval greater than 6 months) compared to “platinum-resistant” disease (relapse free interval less than 6 months). Tumors from patients with “platinum sensitive” disease are indeed significantly more platinum-sensitive in vitro in cell culture assays with cell death enpoints than are tumors from patients with “platinum-resistant” disease. I would think that other labs performing ITRT in ovarian cancer would have similar data, as well.
5. Striking correlations with known patterns of disease-specific activity: e.g. cisplatin being much more active in untreated ovarian cancer than in any other adenocarcinoma and being essentially inactive in non-Hodgkin’s lymphoma, renal cell carcinoma, and carcinoid/Islet cell tumors. Oxaliplatin being equally active in colon cancer as in ovarian cancer and vastly more active in non-Hodgkin’s lymphoma than cisplatin.
6. Cisplatin and carboplatin being almost perfectly cross resistant, while oxaliplatin being only partially cross-resistant, again, with disease-specific activity patterns being consistent with known clinical experience.
I think that putting all these data together in a single paper would provide compelling evidence for the biological validity of ITRT for the most important class of drugs in cancer chemotherapy.
- Larry Weisenthal/Huntington Beach, CA USA
I had the opportunity to attend this session at the AACR meeting in Denver and I provide a few comments:
The study conducted by TGEN, TBAC and Claris MPI (?) included 9 centers in the US and used immunohistochemical and microarray platforms to identify “targets” for drug therapy in a population of previously treated patients.
The algorithm compared each patient’s last TTP (A) as the denominator and their “targeted” TTP (B) as the numerator to identify a 1.3 fold improvement i.e. B/A = 1.3.
The null hypothesis was 15% or less of the sample would achieve this 1.3 fold improvement with a 90% power, P = .05. Drug selection was hierarchical — with gene array + IHC, followed by IHC alone, followed by gene array alone, with pts failing to provide a result offered “physicians choice.”
Patients signed informed consent, all required a biopsy, needed to be off active therapy for 6 weeks and needed to be resistant to prior therapy.
106 patients enrolled
86 (81%) had successful profiles
20 (19%) did not have successful profiles
Of the 86, 18 did not go on to receive targeted therapy (usually because of progressive disease), and 68 were offered targeted therapy with 66 actually treated with targeted therapy.
Breast 18; Colon 11; ovarian 5; other 32
53% of the pts had received 2-4 prior therapies
38% had received 5-13 prior therapies
9 % had received 1 prior therapy
A “target” was identified in 100% of the 66 patients.
Results:
18/66 pts achieved the 1.3 fold improvement (p = .007); median 2.3, range 1.3 to 8.15
Objective response rate was 10%
By disease: 39% of breast, 36% of colon, 20% of ovarian and 6% of others achieved the 1.3 endpoint.
Median OS for the 66 patients treated with assay-directed therapy was 9.7 months versus the OS for all 106 patients of 5 months (but this included the patients who had progressive disease and deteriorated before any treatment at all could be given).
The reason patients did not proceed to therapy was disease progression in 75%.
A few comments:
Pros:
Establishes the willingness of physicians to use “directed” therapies
Establishes the feasibility for conducting these trials
Establishes a criteria for study design with other platforms
Puts assay directed therapy into common oncologic parlance
Cons:
Only 62 % of all the accrued patients were treated.
The drop out of 40 patients may have served to cull the most vigorous from the herd — introducing a bias for survival regardless of intervention.
Most importantly, the simple “identification” of a mutationally activated target i.e. EGFR and ultimate patient outcome with an EGFR antagonist are only loosely related with GBM (vIII mutation) vs. NSCLC (codon 19-21 mutation) providing very different response rates as only one of innumerable examples.
Robert Nagourney (Long Beach, CA)
I submitted to the Committee on Comparative Effectiveness Research Priorities, who is conducting a study to recommend national priorities for comparative effectiveness research conducted or supported with funds from the American Recovery and Reinvestment Act of 2009, a specific priority via their survey tool.
I asked that they compare the effectiveness of various genetic and cell culture assay technologies for targeted as well as conventional cancer treatments. The purpose of the study is to show what technologies work.
The primary condition to be studied would be oncology and hematology. The types of interventions to be compared would be testing, monitoring and evaluation (e.g. lab, functional assessments), pharmacological treatments, and treatment pathways (e.g. chemotherapy selection).
The types of research that would be most effective in providing the needed evidence would be synthesis of existing evidence (e.g. qualitative review, meta-analysis), primary research using existing health care databases, primary research using prospective data collection without randomization (e.g. observational study, registry), and primary research through a prospective randomized trial.
Real-world studies are not being performed under real-world conditions. No one is publishing real-world studies, except private laboratories performing cell-based analysis, which can only do real-world studies, because their studies require fresh, viable specimen, which must be accessioned and tested in real-time under real-world conditions.
Patient outcomes need to be reported in real-time, so patients and physicians can learn immediately if and how patients are benefiting from new diagnostics and therapies.
Greg Pawelski said:
That’s terrific, Greg. Here’s hoping that someone pays attention. It would be terrific to have a head to head comparison of “genetic” and cell culture technologies.
- Larry Weisenthal/Huntington Beach, CA
Additional comments on Von Hoff AACR study (from Medscape, requires easy registration):
http://www.medscape.com/viewarticle/701690?
Push is on to tailor cancer care to tumor’s genes
The whole concept of proper genetic markers is not to put patients in the position of having to receive toxic cancer drugs if they’re not going to do any good. However, genomics is far too limited in scope to encompass the vagaries and complexities of human cancer biology.
The situation with Erbitux and Vectibix for colon cancer, Iressa and Tarceva for lung cancer, and Herceptin for breast cancer is that all the mutation or amplication studies can tell us is whether or not the cancer cells are potentially susceptible to this mechanism of attack.
They don’t tell you if one drug or the other is worse or better than some other drug which may target this. The cell is a system, an integrated, interacting network of genes, proteins and other cellular constituents that produce functions.
No genetic profile can discriminate differing levels of anti-tumor activity occurring among different targeted therapy drugs. Nor can it identify situations in which it is advantageous to combine a targeted drug with other types of conventional cancer drugs.
“Targeted” drugs are poorly-predicted by measuring the ostansible “target,” but can be well-predicted by measuring the effect of a drug on the function of live cells, the net effect of all processes, not just the individual molecular targets.
The benefits of newer targeted therapies are marginal. These targeted therapies may impart a clinical benefit by stabilizing tumors, rather than shrinking them (substituting shrinkage for stabilization).
I would not want to be denied treatment with any targeted therapy because of a gene mutation or amplication. Genetic testing is not a clear predictor of a lack of benefit.