Oncology- Genetic Testing, AI, and Personalized Medicine

Written By Harshil Shah

Author: Sandy Lum

TLDR

Genetic Testing- Invitae Corp and others offer more affordable and accessible genetic testing

Artificial Intelligence- Skin Vision uses AI to detect signs of the most common types of skin cancer

Personalized Radiation Therapy- Varian uses the day-of anatomy and positioning to adjust radiation treatment plans

Genetic testing for cancer risk: cheaper and more accessible

A core principle of cancer screening is to allow the early detection of cancer while a patient is still asymptomatic, to allow prompt therapeutic intervention that can not only reduce mortality rates from that cancer, but can also enable treatment modalities which generally are less aggressive and have fewer side effects for the patient. Genetic testing, which investigates the DNA sequence or chromosomal structure of a patient to identify genetic mutations, can be used to help profile that patient's individual cancer risk. While testing positive for a genetic mutation does not definitely determine whether a patient will develop the illness, it can inform critical health decision-making as well as promote awareness and preventive behaviors. A previous longitudinal study of direct-to-consumer personal genomic testing found that undergoing this service was associated with primarily positive changes in diet and physical activity. 

 Genetic testing has been well-established in clinical practice with healthcare providers for some time, however barriers such as limited access to care, high cost, and the challenge of contextualizing the data have limited the potential for genetic screening to improve health outcomes. Thus, any innovation to genetic testing that would increase accessibility, decrease cost of sequencing and administration, and support meaningful integration of genetic information into health decision-making can make a significant impact.

 Invitae Corp. is a biotech company that has honed their efforts on reducing the barriers to genetic testing with a mission of offering more affordable and more accessible tests. In theory, a wide and routine adoption of genetic testing by patients and doctors alike would accommodate lower test costs. Invitae offers direct-to-consumer and -physician genetic testing that attempts to reduce barriers to testing by allowing patients to request tests online, provide saliva samples from home, and review results online, at costs that are generally less expensive than genetic testing obtained through a healthcare provider. 

Artificial Intelligence in cancer diagnostics

Artificial intelligence (AI)  is becoming rapidly prevalent in many branches of society, including healthcare. The field of oncology moves at an accelerated pace of breakthroughs in research resulting in advancements in care, so it is no surprise that AI has significant potential in this branch of medicine. 

Hunter et al have discussed how AI can analyze a multitude of existing data such as (1) electronic healthcare records, (2) radiology images, (3) digital pathology slides, and (4) multi-omic data to improve early cancer diagnosis in a myriad of applications: (1) stratified screening, (2) symptomatic patient triage, (3) workflow triage, (4) image analysis, (5) biomarker analysis, and (6) recurrence detection/prediction. 

It must be noted that AI in cancer diagnostics is not just theoretical. Researchers around the world are developing and clinically validating AI solutions to help clinicians make earlier and more accurate diagnoses - take for example, NYU Langone’s Perlmutter Cancer Center’s free online AI tool which analyzes and classifies central nervous system tumours. This tool was approved by the state for use as a diagnostic test in 2019.

Kheiron Medical Technologies is a company that develops deep-learning solutions to improve imaging workflows for radiologists. Their commercial solution, Mia, focuses on helping radiologists with decision-making in breast cancer screening to independently identify abnormalities, reduce clinical loads, and reduce patient wait times. 
SkinVision is an example of a medical service that can be directly utilized by patients; it utilizes AI algorithms with the goal of helping patients detect potential cancer early. Users can download the SkinVision app, a regulated medical device that uses AI to detect signs of the most common types of skin cancer. If any risk is identified, the user is advised to consult with a doctor to make a definite diagnosis.

Personalized radiation therapy

Radiotherapy is a form of cancer treatment that uses high doses of ionizing radiation, typically generated by linear accelerators, to cure or control cancer cells. Patients may receive radiotherapy alone as their cancer treatment, or in conjunction with surgery and/or systemic therapy. Radiotherapy has been applied in cancer treatment since the late 19th century, but the approach has radically changed over the years. Radiotherapy once utilized broad, generic radiation treatment field that encompassed the patient’s tumour as well as normal tissue; now the treatment is much more personalized to the patient - multi-leaf collimators dynamically shape radiation beams in real time based on the patient’s unique tumour shape as the treatment head moves in an arc around the patient while also dynamically modulating the radiation beam size and intensity. 

The latest radiotherapy machines account for changes in a patient’s anatomy, which for some can change significantly from factors such as weight loss during the course of a treatment (some courses of treatments span up to 7 weeks). Elekta is a company that offers treatment machines that not only deliver radiation therapy but also takes MRI scans as the treatment session progresses, allowing the radiation oncology team to see the treatment area in real time. 

Radiation oncologists at Columbia University Irving Medical Center utilizes adaptive online radiation therapy. This latest advance utilizes a CT scanner built into the treatment machine, which scans the patient to identify anatomical changes before each treatment. Rather than requiring clinical staff to delay therapy after significant anatomical changes to create a new plan, a machine learning module automatically uses the CT scan’s 3D image to recalculate the plan which takes the radiation therapy team only a few additional minutes to review and approve.

Varian is one example of a company that makes linear accelerators that utilize AI in this same manner. Their Ethos machine is an AI-driven radiation therapy solution that uses a patient’s day-of anatomy and positioning to adjust the radiation treatment plan to offer highly personalized, precise care without treatment delay.

References:

 1. Nielsen DE, Carere DA, Wang C, Roberts JS, Green RC; PGen Study Group. Diet and exercise changes following direct-to-consumer personal genomic testing. BMC Med Genomics. 2017;10(1):24. Published 2017 May 2. doi:10.1186/s12920-017-0258-1

2. Hunter B, Hindocha S, Lee RW. The Role of Artificial Intelligence in Early Cancer Diagnosis. Cancers (Basel). 2022;14(6):1524. Published 2022 Mar 16. doi:10.3390/cancers14061524

3. Capper, D., Jones, D., Sill, M. et al. DNA methylation-based classification of central nervous system tumours. Nature 555, 469–474 (2018). https://doi.org/10.1038/nature26000

4. https://www.cuimc.columbia.edu/news/smarter-radiation-therapy-enhances-personalized-care-columbia-and-newyork-presbyterian-cancer-patients

Further Reading:

Genetic testing for cancer risk - cheaper and more accessible

Harshil Shah
Previous

ENT- VR/AR, 3D Printing, Digital tools and Wearables
Next

Orthopedics- VR, Implants, and Regeneration