The Meaning of Meaningful Improvement in Lung Cancer

When asked to define what constituted pornography in his 1964 Supreme Court decision (Jacobellis versus Ohio 1964) Justice Potter Stewart stated, “I know it when I see it.” When I reviewed an article on the changing landscape of clinical trials in non-small cell lung cancer (NSCLC) (Shifting patterns in the interpretation of phase 3 clinical trial outcomes in advanced non-small cell lung cancer: The bar is dropping, Sacher A. G. et al, J Clin Oncol May 10, 2014), Justice Stewart came to mind.

The authors selected 203 NSCLC trials from a total of 245 studies conducted between 1980 and 2010. They compared how often the studies met their endpoints with how often the study authors’ called the results “positive.” Among the findings, it seems that earlier studies (before the year 2000) were geared for overall survival, while later studies (after 2000) overwhelmingly favored progression free survival. Although patient survivals changed little, the number of trials reported as successful increased dramatically.

Non-small cell lung cancer

Progression-free survival measures how long it takes for a patient to fail treatment. That is, for the disease to worsen on therapy. Its use increased after 2000 when Docetaxel, for the first time, provided a survival advantage in recurrent disease.

The FDA’s willingness to accept progression-free survival for drug approval was originally based on their expectation that the benefit would be “substantial and robust” but they did not define the term. One group has suggested that improvements should be of the magnitude of 50 percent. Another went even further suggesting a doubling of the survival advantage.

Unfortunately, the trend has been just the opposite. Trials from the 1980s on average gave a 3.9 month improvement, which fell to a meager 0.9 months after 2000.

What are patients and their physicians to make of these trends? First, the large clinical trials, that are so common today, are much more likely to achieve significance. The troubling corollary is that statistical significance is not the same as clinical relevance. The “publish or perish” climate, combined with the skyrocketing cost of drug development has placed inordinate demands upon investigators and their sponsors to achieve “positive results.” Fearing failure, many pharmaceutical companies sponsor “safe” trials that provide incremental advances but few breakthroughs.

Meaningful advances in oncology are generally quite evident. The first use of Interferon alpha for the treatment of hairy cell leukemia provided a response rate of 100 percent and earned a lead article in the New England Journal of Medicine (NEJM) with only seven patients!

Similarly the 57 percent response rate for Crizotinib in ALK positive lung cancer required only 82 patients for a place in the NEJM. Unfortunately, the failure of contemporary investigators to identify more “paradigm changing therapies” has forced many to lower the bar.

The clear solution to the problem is the better selection of candidates for therapy. Despite advances in molecular biopsy a paucity of truly effective companion diagnostics exist. Outside of EGFR, ALK, and ROS-1, it is anybody’s guess how to manage the vast majority of non-small cell lung cancer patients.

While we expand our armamentarium and develop better companion diagnostics, today we can apply measures of cellular response (as found in an EVA-PCD assay)
that capture all of operative mechanisms of sensitivity for all classes of drugs. While it is not always possible to know why a patient will respond, it is possible to know that they will respond. In the words of Judge Stewart, when it comes to a responsive lung cancer patient “I know it when I see it.”

Every Cancer Patient’s Outcome is Important

Clinical oncologists can be divided into different camps. There are those who see patient outcomes as a means-to-an-end. Each clinical response provides a data point and when those data points reach critical mass they become reportable. These are the trialists. They see the world through a utilitarian lens. They use aggregate data, through sufficient patient accrual, to achieve significance. This, they hope, will lead practice-changing observations. Trialists populate academic centers and an ever-expanding number of “mega medical groups” that are now gobbling up private oncology practices. They apply metrics to gauge success, as their focus has moved away from individual patient needs toward the achievement of a “greater good” for the population as a whole. Statistical significance is the currency of their realm and clinical protocols their preferred tool.

In the other camp reside physicians, that dwindling cadre of doctors whose principal focus is the good response of each individual patient. They are the practitioners who eke out a living in an environment of diminishing returns. Having relinquished both autonomy and income over recent years, their one remaining reward is the benefit they can bring to each patient. With neither the desire nor ability to publish their results, individual patient survival becomes their paramount goal. Their job is to alleviate suffering, provide comfort and sponsor the health of their clients. Patients preparing to meet with a cancer specialist should consider carefully who is treating them – and why.

I was reminded of this when a 48-year-old gentleman recently requested an opinion. He had presented to an emergency room with a month-long history of sharp abdominal pain. The CT scan revealed extensive intra-abdominal disease, which upon endoscopic biopsy, proved to be of gastric (stomach) origin. He was immediately referred to an accomplished university-based clinical investigator for consultation.

Metastatic gastric cancer is a very difficult disease to treat. One bright spot has been the discovery that 20 percent of patients carry an epidermal growth factor receptor (HER-2) mutation that enables them to receive Herceptin-based therapy. As luck would have it, this patient did not carry the HER-2 mutation. The university investigator explained that there were limited treatment options. In light of his metastatic presentation, the doctor felt that aggressive, multi-agent chemotherapy might only engender toxicity. The patient was offered either single agent 5-FU for palliation or the opportunity to participate in a clinical trial. The patient considered his options and chose to seek an opinion with me.

I reviewed the patient’s status and explained that while the opinion of the university investigator was valid it might underestimate the patient’s individual chance of response. I explained that gastric cancer statistics, like all medical statistics, are population based. That is, a 20 percent response rate does not mean that every patient gets 20 percent better, but instead, that 20 out of every 100 respond while 80 do not. Our job was to find out which group he belonged to.

The patient decided to undergo a biopsy and submitted tissue to Rational Therapeutics for EVA-PCD® analysis. The results were strikingly favorable with several drug combinations revealing both activity and synergy. After careful comparison, I recommended the combination of a Cisplatin, Taxotere and 5-FU (DCF), a regimen originally developed at the MD Anderson almost 10 years earlier.

On March 12th, the patient began treatment on an every-other-week schedule. As he did not circulate tumor markers like CEA or CA 19-9, there was no easy measure of his response so I elected to repeat the PET/CT after just two cycles. Much to my delight, the patient had achieved a complete remission with resolution of all measurable disease, including the bulky abdominal masses, numerous lymph nodes and the stomach. As I described the remarkable PET/CT results, the patient’s wife began to weep. Her husband, the father of their two young children, wasn’t dying after all. He was no longer a grim cancer statistic. With mother’s day approaching, this was the first good news that they had received in six months. At once, the patient began to discuss business trips, travel plans and family outings. He breathed a slow sigh of relief as he realized that, once again, he had a life.

Good outcomes, even in the worst diseases, occur in all oncology practices. Every doctor can regale you with the story of a patient who responded beautifully and went on to survive years beyond everyone’s expectations. The reason we remember these stories is because they occur so infrequently. Complete remissions in metastatic gastric cancer are vanishingly rare. That is the reason that the university investigator offered single agent 5-FU. It’s easy, nontoxic, well tolerated, but it also cures no one. The rationale is well established: Why poison patient’s you cannot cure? Playing the averages, this strategy is a winner. Yet, on an individual patient basis it may, in fact, be a very big loser.

What are we to do with the “non-average” patient? What about the outliers? Should we not, at least, try to find them? We do it with stocks, racehorses, Indy-drivers, real-estate investments and every underdog sports team in every league. It’s the outliers after all that we call winners.

Cancer patients are not clinical trial subjects. They are unique individuals with their own very unique biology. Every patient is an experiment in real time, an “N of 1.” We must respect the dignity of each individual and we are duty-bound to apply every tool at our disposal to assist him or her in the pursuit of his or her own very personal best outcome – providing truly personalized cancer treatment. This patient did not have a 20 percent response. Instead, he was one of the fortunate few who responded very well. And for him that response was 100 percent.

Cancer Patient’s Need to Know – Whose Avatar is it?

The 1984 celebrated case of Baby Fae described the efforts of intrepid investigators at Loma Linda University in California to save the life of an infant born with a hypoplastic heart (the left side of the heart was severely underdeveloped). To salvage this unfortunate child, the heart of a baboon was transplanted into her chest, which successfully maintained her cardiac function. Unfortunately, the child died 21 days later after rejecting the heart.

The experience nonetheless spawned a lively discussion of ethics in human experimentation and the lengths to which we will go to save a life. A friend residing in New York contacted me shortly thereafter. Knowing that I was in medical training in California, he was intrigued by this report and posed the question. “What is the fastest animal in the world?” When I suggested a cheetah or African gazelle, he answered “No, a baboon running past Loma Linda Medical Center.”

I am reminded of this quip following a news report about Champions Oncology. This biotech company located in Hackensack, NJ, founded by investigators from Johns Hopkins University has developed a technique to transplant human tumors into immunocompromised (known as nude for their loss of hair) mice to test drugs and combinations. They use the term “avatar” to describe these human-tumor-bearing mice and like their video-game counterparts these mice serve as surrogates for their patient “users.” It takes 20 to 30 mice to complete the analyses for each patient. Although the work is based on sound science, the practicality, predictive validity and ultimate utility of this approach has yet to be established. This has not dampened the enthusiasm of desperate cancer patients who have proven willing to spend tens of thousands of dollars to undergo a Champion analysis.

So what are the upsides of the avatar model? For one, this is a living organism with a functioning blood supply, liver, kidneys and the capability of metabolizing pro-drugs (precursor) into active species. The process utilizes cell clusters, not individual single cells in their analysis. Responses are ultimately “phenotypic” in as much as they reflect cellular responses to injury and not genotypic profiles. Finally, toxicities can also be assessed by measuring the animal’s tolerance of the drug or combination administered.

But let’s drill down for a moment and take a closer look. The avatar approach requires months of preparation, the operation of a vivarium (animal zoo) and several mice for every single drug or combination tested. It requires prolonged (many months) maintenance of the animals with highly unpredictable engraftment of the transplanted tumors. Furthermore, significant time, energy and skill are needed to maintain these in-vivo systems.

Ultimately, only a handful of drugs can be examined, lest the number of mice required becomes unmanageable. At the end of the day these investigators are making a valiant effort to approximate work that we, and our colleagues, have successfully conducted for more than two decades – the accurate selection of chemotherapies and drug combinations for individual patients.

If we allow for the obvious downsides of expense, difficulty, time, limited sample size, low efficiency and resource intensity required to conduct even a single patient’s study there are more daunting concerns.

First, these cells are not actually in their “native state.” Over a period of time the tumors will no longer be host to human immune cells, nor will they be exposed to human cytokines and VEGF. The observed growth of the implanted tumors in the mice-avatars may, in part, reflect an ingrowth of mouse-derived fibroblasts and blood vessels, which have distinctly different biology from those of a human host.

Even if we accept the expense and difficultly of avatars, there is no clear evidence on an individual patient basis, that this approach holds any advantage over the much simpler and direct evaluation of human primary culture microspheroids. That is, the avatar approach appears to be a difficult, cumbersome, inefficient and a very expensive way to do something that we can already do inexpensively, rapidly and efficiently. Further, the purported advantages of in vivo-avatar system are actually less than meets the eye.

After all, most clinical drugs have “active” derivatives that can be utilized for testing in short-term culture without the need for a mouse liver. Our careful calibration of in vitro drug exposures against actual patient responses (P < 0.001) has established the predictive validity of these culture conditions. Finally, the toxicities of virtually all clinically relevant drugs that patients would request for testing (and likely receive) are already well-known to clinicians from existing Phase I and Phase II clinical data sets.

Human tumor sensitivity to chemotherapy (or targeted agents) is driven by what might be described as “response elements.” These unique features of each patient’s tumor can be accurately probed at the phenotypic level through the use of simple assays conducted in short-term culture. Our microspheroid model has proven highly predictive of clinical outcome in virtually every tumor type ever tested.

Our analyses are conducted in seven days, with samples that are the same size or smaller than those required for avatar generation. Furthermore, short-term platforms can analyze dozens of drugs and combinations at a price that is far less expensive. While it might be argued that avatars, once established, can be used as repositories for future research, that is small comfort to patients in need of immediate answers who find themselves paying handsomely for a service that will not be available in a timely manner, e.g. that which can help them in their need for immediate drug selection.

It seems that the medical science community is less interested in results than process. The fact that short-term cultures are predictive of clinical outcome seems less important than the provocative scientific results that these avatar models can provide Avatars enable scientists to interrogate cancer cells for genomic and proteomic signals, offering the opportunity to conduct interesting science. But has that science become more important than the clinical utility of the tests that were purportedly developed (and sold) to improve patient outcomes?

Patients who are considering spending tens of thousands of dollars for these glorified chemosensitivity tests would do themselves a service to first carefully examine the predictive validity, breadth of data, cost and turn-around-time of short-term culture methods, like the EVA-PCD® assay before they commit their precious time and resources to so “interesting” an endeavor as an avatar analysis. After all, it is the patient and their good outcome that should be at the top of the list when the advantages of any system or method are being weighed. Truly personalized cancer care should be just that – personalized.

With the rise of avatars it may be timely to re-examine the original question and wonder whether the fastest animal in the world will soon be a nude mouse running past Johns Hopkins University.