Cancer Centers Hitting the Target

We Are Precise, But Are We Accurate?

Dan S. Zuckerman, MD, FASCO


J Oncol Pract. 2019;15(6):305-306. 

"I used to bullseye womp rats in my T-16 back home."
Luke Skywalker

Luke uses the force to guide his X-wing fighter through the trenches at the end of Star Wars. A perfect hit to the thermal exhaust port will lead to a chain reaction that will destroy the Death Star and save the galaxy. Precision and accuracy—the dream of every kid of my generation and every oncologist of the post-imatinib era.

Levit et al[1] describe three different models addressing the implementation of precision medicine in a well-recognized integrated community health system (Intermountain), a large hybrid academic-community center (Levine Cancer Institute), and a consortium of independent oncology practices from across the country (National Cancer Care Alliance [NCCA]). The challenges are shared among all community oncologists practicing in the early 21st century: the intrinsic complexity and unknowns of genomic/proteomic medicine, the dearth of basic science and translational investigators in community centers, the logistics of setting up molecular tumor boards (MTBs), the stratospheric expectations of patients and families, and the financial barriers and burdens of paying for off-label targeted therapy.

Levit et al[1] and their respective institutions should be lauded for developing structure around what is otherwise a Wild West of molecular medicine. Their article provides potential models for other centers to aspire toward, while simultaneously raising questions about how much time, expertise, and money it would take to get there. Many practices are struggling with questions, such as: Should my institution embark on in-house next generation sequencing (NGS)? How do I set up an MTB, and what are the criteria for establishing it as an expert panel? Will I be able to fund a navigator for clinical trial matching or bird-dogging insurers for approval of off-label drugs?

It takes a special community cancer center, however, to set up an in-house precision medicine program. At St Luke's Mountain States Tumor Institute, a not-for-profit encompassing five regional cancer centers and seeing 3,500 new patients with cancer annually, we have entertained the idea of developing an in-house NGS platform. Five years ago, we ordered NGS sparingly, and now we hardly see a patient with advanced solid cancer who does not get NGS. Scientific curiosity, an ambitious pathology colleague, and the oncologic version of keeping-up-with-the-Joneses as we watched our regional cousins Swedish Cancer Institute in Seattle and Intermountain in Utah develop their own NGS programs have prompted us to consider setting up such a program. In meeting with my colleagues, I have asked: "Why would we do that when we can just order the test from Company X?" Meeting with my administrator: "A sequencer costs how much?!?" Interview with a bioinformaticist PhD candidate (eyes glazed, scratching my head): "… and QC pass rate … needs hybrid capture bait specificity … terabytes… ." To date, St Luke's has not set up in-house NGS testing because there are still no comparative data to guide us that one center's (community or academic) boutique NGS is better than some other commercial NGS platform, and vice versa. So, we order NGS from Company X and proceed with what was referred to by Levit et al[1] as ad hoc decision making.

St Luke's has also struggled to justify the effort and expense of implementing MTBs, as was done by all three models referring to MTBs, with Levine Cancer Institute holding them weekly, Intermountain biweekly, and NCCA at some of its member practices but not others. It would be great to know from the authors who staff their MTBs: pure academicians from NCCN centers, MD/PhDs, or everyday medical oncologists? Even putting credentials and basic science chops aside, what are the criteria by which one gene alteration is ranked as more actionable than another? Let's say an NGS report finds mutations MAP2K2, NOTCH1, PTEN, and RET and is accompanied by an MTB interpretation that reads, "suggest the following potential therapeutic interventions in the ranked list below: RET and PTEN." What data inform this interpretation?

MTBs present a logistic challenge as well. In 2016, with the help of ASCO and the Idaho Society of Clinical Oncology, St Luke's established the first statewide virtual tumor board in the country. It was held monthly, with video conferencing allowing any oncologist in the state to call in, and with scientific interpretation provided by experts from the Seattle Cancer Care Alliance. It was new. It was exciting. I learned more about genomic pathways and their putative role in driving this cancer or that. However, we were unable to continue past one year because of spotty physician participation, lack of clinical timeliness for patients, and unsustainable reliance on an outside academic center.

Another area of difference between the three programs highlighted and many community cancer centers is the expertise of the staff. NCCA's model is the least developed, but it is also probably the most representative of the average community practice across the country, my own included. This should not be surprising. Intermountain stands out for innovative leadership and benefits from being one of the few community centers in the country with MD/PhDs who could develop and oversee an in-house NGS program. Levine Cancer Institute is one of the largest health systems in the United States, and it is cross-pollinated with academics.

It would also be useful to know how these precision medicine programs were funded, particularly if there were arrangements with pharma, an outside NGS vendor, and/or philanthropic contributions. Many features they described are not found as line items in the typical community cancer center budget. MTBs and clinical trials navigators are labor and time intensive and seem unlikely to be reimbursable. Intermountain administers their own health plan, SelectHealth, creating a financial incentive to in-house its NGS in a way that may not exist for other centers. NCCA references access to genomic testing for their network, which leaves us wondering if they standardized their NGS vendor across their practices and, if so, if they were able to leverage some purchasing power in this.

Overall, the authors show us that we can set up precision medicine programs at our own centers. But they do not answer whether we should. To date, we lack any definitive data that the wholesale approach of precision medicine actually improves patient outcomes. The SHIVA study in 2015 showed no difference when comparing it to standard-of-care chemotherapy. We await the results of the NCI-MATCH and ASCO's TAPUR trials with bated breath. We have well-validated studies demonstrating the superiority of specific agents targeted to canonical mutations like EGFR and ALK in non–small-cell lung cancer and BRAF in melanoma, but we have yet to prove that widespread adoption of NGS platforms, as enticing as it is, will translate to equally widespread improvement in survival.

As Luke Skywalker's X-wing fighter sped through the trenches, he and his fellow rebels had the data on which thermal exhaust port to target and that a direct hit would lead to the destruction of the Death Star. Are we sure that our precision medicine is as accurate as Luke? I suspect we will have to wait a few more episodes to find out.