By Johannes E. Wolff, MD
Cleveland Clinic is a non-profit academic medical center. Advertising on our site helps support our mission. We do not endorse non-Cleveland Clinic products or services Policy
We practice in an era that can easily create ambivalence for the pediatric oncologist. Never before have so many novel therapeutic agents for cancer been available, yet their safety in children is largely unknown, since most toxicity studies have been conducted only in adults. Moreover, these agents’ efficacy has been demonstrated only for common cancers for which it was possible to assemble large populations for phase 3 clinical trials. This leaves pediatric oncologists with a handful of tried-and-true agents used in highly standardized protocols developed from multi-institutional consortia — but with precious little guidance for treating rarer cancers or treatment-refractory cases.
The case for targeted therapy
Today, pediatric oncologists at Cleveland Clinic Children’s are taking a novel approach by offering targeted therapy to children with cancers unresponsive to standard treatment options. In contrast to strategies based on traditional histological diagnosis, this promising form of individualized therapy involves analyzing each tumor with genetic and morphoproteomic methods using a panel of markers to reveal treatment targets. Based on this analysis, a drug is selected from a list of novel agents known to be effective against these markers.
The challenge — and opportunity — of too much information
Analysis of individual tumor composition requires examination of a plethora of factors, including genetic analyses, protein expression, pathway activation in tumor cells, spatial tumor heterogeneity and tumor development over time. Digesting this enormity of information and turning it into actionable clinical decisions is highly time-consuming and often overwhelming. Moreover, reports from different laboratories typically do not agree, and some seem to suggest no actionable items at all.
Yet on the flip side of this challenge is an opportunity. The opportunity lies in channeling the individual pediatric oncologist’s experience and knowledge into a multidisciplinary discussion to identify — with support from innovative technology and in the setting of a targeted therapy tumor board — the medication(s) most likely to benefit an individual pediatric patient. At Cleveland Clinic Children’s, we are taking this opportunity by partnering with adult-care oncologists and pharmacists in Cleveland Clinic’s Taussig Cancer Institute in an elaborate, standardized process that supports selection of the most appropriate agents.
Ensuring patient safety in a changed landscape
The agents most commonly considered for tumor targeting are FDA-approved for other diagnoses, thus requiring off-label use in pediatric cancers. Whereas in the past we could readily form nationwide committees to develop age- and weight-appropriate dosing and ensure safety for off-label pediatric uses, this solution is not consistently applicable today, with 10 times as many agents available, less common malignancies to be treated and a need to make individual therapy decisions more quickly.
To ensure patient safety in this changed landscape, Cleveland Clinic Children’s is employing the latest information technology to monitor for known drug interactions and other safety red flags. Doses are escalated cautiously. Clinical monitoring is extremely vigilant, with laboratory, imaging and functional tests conducted far more frequently than for patients undergoing traditional standardized treatments with one protocol for all.
Patient by patient, we have built our targeted therapy experience into a broadly standardized approach. Patients on targeted therapy are initially examined twice weekly and closely monitored with electrocardiograms and blood, liver and kidney function tests even in the absence of clinical signs. Under this surveillance, doses are gradually increased and synergistic combinations developed until we are confident their anti-cancer effect is not matched by additive toxicity. Once the patient’s therapy schedule and dosing are stable and have proved tolerable, monitoring can be eased.
Harnessing statistics to predict success
Determining whether a new therapy choice works requires treating the patient, collecting data and analyzing the outcomes. It takes many attempts to learn how to do it right.
Our treatment decisions as clinicians are generally based on what has worked before — or what should work, given the biology of a particular tumor. However, tumors of the same type often differ.
The dead cancer cells seen under a microscope might differ from growing cancer cells. Two patients with the same diagnosis may differ from each other. Even the cancer cells within one individual might differ in different body locations — or may change over time. For these and other reasons, a treatment protocol might succeed in one patient and fail in another.
This argues for an individualized approach in which the standard for determining efficacy — the phase 3 trial requiring a large sample size — is simply not possible. This reality, together with the volume of tumor information available and the clinical challenges of giving novel drug combinations, demands enhanced use of biostatistics to predict and determine treatment efficacy.
Several new statistical tools have been proposed. Our targeted therapy program uses an approach developed at MD Anderson Cancer Center, where I practiced before coming to Cleveland Clinic Children’s. At the model’s core is the comparison of an individual outcome prediction with the observed outcome following targeted therapy. The approach requires a research project description, IRB approval, patient consent and other trappings of traditional clinical studies, but it combines them with a number of more novel features:
- Use of large amounts of data to predict individual outcome
- Mathematical models to maximize predictive accuracy
- Quantification of differences between predicted and observed outcomes
- Hypothesis-driven analysis of the quantifiers used
Figure. Schematic showing the central role of data-driven artificial intelligence in the process to continuously improve targeted therapy treatment selection for pediatric malignancies.
These features, which are reflected in the above schematic (Figure), amount to an artificial intelligence-like approach to therapy selection using a biostatistical algorithm that harnesses big-data computing capabilities to improve medication choice.
Next step: networked data analysis
While we are excited by this new data- and biostatistics-driven targeted therapy paradigm at Cleveland Clinic Children’s, its utility is limited while it remains at the scale of a single institution. We do not plan on doing this work alone, and the program is now ready for the next step: networking.
Past successes in pediatric oncology have stemmed from combining the efforts of multiple institutions. Targeted therapy will follow the same path. After all, the main challenge for targeted therapy in pediatrics is no longer tumor analysis but rather how to analyze treatment outcomes so that cumulative experience from the past will lead us all to better treatment decisions in the future.
Standardized, ongoing data collection is the key to proving the efficacy of targeted therapy and eventually making it practical, affordable, and capable of continuous refinement and improvement. For this effort, all hands will be needed on deck, and we welcome the opportunity to move forward in collaboration with pediatric oncology colleagues around the nation.
Dr. Wolff is Chair of Cleveland Clinic Children’s Department of Pediatric Hematology, Oncology and Blood & Marrow Transplantation.