Sharing

Image source

I found the following while monitoring my curated Twitter feed for this project (h/t Tom Silver, ‏@TomSilver39):

"Does anyone have early access to inside information?  There is another way that CytRx’s clinical results are made available to big pharmaceutical companies, and that is CytRx’s “black box” website. This website is designed to provide big pharma with inside information, with the objective of interesting them in possible partnerships, which can enhance shareholder value by providing cash to accelerate and/or expand the pipeline without selling more shares.  Big pharma companies who wish to monitor research results as they occur, can sign a nondisclosure agreement, and are then granted access to this restricted site. There, they can view results at almost the same time that CytRx receives them from the clinical testing centers.  Now, if it is a double-blind study, they will still not know which patients received Aldox, and which received the control drug(s) or placebo. Still, it is not difficult to calculate the therapeutic benefit of Aldox, using known historical data on the control drug, and the ratio by which patients are assigned to each arm of the trial. For example, suppose that Aldox is being compared to Dox, and every 3 patients are randomly assigned in a 2:1 ratio Aldox:Dox. If Dox gives a progression-free survival of 10 months, and the average PFS for all of the patients is 20 months, we can calculate that the PFS for Aldox is 25 months because (25 + 25 + 10) /3 = 20." Source link: CytRx Corporation Overlooked News {Bold emphasis is mine}

There are two topics here. First, there is subject of collecting clinical trial data in real-time, which should come as no surprise to anyone (given the advent of information technology now on its umpteenth wave of progress). It also goes without saying that if you can collect data in real-time, or as clinical sites send these data in, you also can analyze and share it. A sample paper on the topic: Harnessing technology to improve clinical trials: study of real-time informatics to collect data, toxicities, image response assessments, and patient-reported outcomes in a phase II clinical trial, Pietanza et al., J Clin Oncol. 2013 Jun 1;31(16):2004-9. doi: 10.1200/JCO.2012.45.8117. Epub 2013 Apr 29.

Second, there is the subject of sharing the data with third parties. See Pharma Compliance Monitor's December 29, 2014 blog post entitled Clinical Trial Data Sharing: Landscape, Trends, and Risks by Robin Jenkins of Sanofi U.S. and Joe Morrell of Huron Life Sciences:

"The potential benefit of sharing clinical trial data (patient-level data and clinical documentation) has been recognized by various industry groups over the past several years, but the reality of sharing data is emerging...[D]ata sharing initiatives can be categorized into three fairly broad data sharing models[1]; Black Box, Gatekeeper, and Open Access."

[1] refers to Mello, JD, PhD, Michelle, “Potential Models for Data Sharing.” Issues and Case Studies in Clinical Trial Data Sharing: Lessons and Solutions; Cambridge, MA, May 17, 2013. The slides below better inform on the black box reference above in the CytRx blog post, which is the name of one of several models of data sharing noted above in the Pharma Compliance Monitor blog post.

Click to enlarge.

Click to enlarge.

Click to enlarge.

The conference above speaks to a much larger topic about leveraging data sharing to accelerate biopharmaceuticals development.

Click to enlarge.

My blog post is more focused on what I learned:

• That all companies running any clinical trial and certain pre-clinical ones can share data with third parties,
• That to gain access to such data, the parties would have to enter into non-disclosure agreements ("NDAs"),
• That software platforms and packages exist to facilitate sharing. INC Research has such a product, and other large global contract research organizations ("CROs") have proprietary ones as well,
• That Big Pharma do enter into NDAs with companies when the former are interested in one of latter's trials, and
• That providing data access to certain melanoma advocacy groups might assist in the effort to bring more visibility to the Stage 3 melanoma patient (progress is gained for constituents when the oncology community is better educated and good public policy is facilitated).

Let's say, merely for the sake of argument of course, that you're Pfizer (or any interested Big Pharma), and you want to look at data generated from Provectus's upcoming pivotal Phase 3 trial for locally advanced cutaneous melanoma. Let's assume:

• The above mentioned process for clinical trial data sharing with third parties is real, appropriate and indeed happens, etc.,
• You sign an NDA with Provectus, and
• Progression-free survival ("PFS") for systemic chemotherapy (dacarbazine or DTIC) is [2-]3 months.

If DTIC gives a progression-free survival of 3 months, and the average PFS for all of the patients is 9 months, we can calculate that the PFS for PV-10 is 12 months because (12 + 12 + 3) /3 = 9. [Randomization is 2-to-1]

Let me frame the analysis another way, using the knowledge gleaned by Eric and Provectus from the compassionate use program ("CUP") and sub-set of the metastatic melanoma Phase 2 trial patients who had all of their disease treated: hit the injectable disease early, often and repetitively until is goes away (i.e., a complete response, or partial response until the complete response is achieved). This approach could (would) suggest a PFS of 1.0 for the pool of suitably treated PV-10 patients (i.e., the PV-10 arm of the upcoming pivotal trial). Let's now assume you re-use the assumption above that the PFS for systemic chemotherapy is 3 months (i.e., the pivotal trial's control arm). A graph of months after treatment vs. PFS in months might look like this:

Click to enlarge.

A graph of months after treatment vs. PFS (as a % or fraction of 1) might look like this:

Click to enlarge.

None of the above graphs are dependent on N, the total number of patients treated at a given point in time. Real-time data collection, or data collection as clinical trial sites provide the data to the CRO running the trial and, thus, the data sharing software sharing package, would simply yield lines, graphs and trends that would change over time as N changes (increases) and the results of treatment on each N changes.

Different thresholds of N would have different meanings for statistical significance and relevance. N-1 might be the number of patients for the divergence of the PV-10 (blue above) and chemo (red above) arms to reach statistical significance. N-2 might be the number of patients for the trial to reach its assigned hazard ratio. N-3 might be the number of patients/responses to trigger the interim analysis.

A Big Pharma with a basic-to-good understanding of PV-10 (e.g., pre-clinical data, prior clinical data, mechanism of action, mechanism of immune response, intralesional therapies, etc.) that also had signed an NDA to access some level of pivotal trial data would observe the trends of the trial's data—positive or negative—before or as the different N thresholds are reached and/or passed.

Then what?