News in brief: Molecular testing highlighted in classification of childhood tumours; Opioid fears a barrier to cancer analgesia adherence; Oncology AI program dumped by IBM

Thursday, 27 Jan 2022


Molecular testing highlighted in classification of childhood tumours

The soon-to-be published inaugural World Health Organization (WHO) Classification of Pediatric Tumors will help improve diagnostic precision and health outcomes in children and adolescents with cancer globally.

A statement from the American Association for Cancer Research said sporadic and hereditary genetic alterations play a key role in the development of paediatric tumours.

The document reflects the transition from diagnoses based on histological/microscopic findings and immunohistochemistry to more reproducible, molecularly-driven diagnosis based on tumour genomics.

“This will additionally require the development of affordable tests and supporting networks for middle- and low-income countries, complemented by artificial intelligence approaches to potentially predict molecular classes from histology samples in the future.”

First author Professor Stefan Pfister, from the Hopp Children’s Cancer Center Heidelberg said “…a precise, unbiased, and unambiguous diagnosis that harmonises molecular tumour typing, prognostic and predictive biomarkers, and potential cancer predisposition is an extremely good investment to improve patient outcomes and spare treatment side effects.”

The Classification of Pediatric Tumors will be published as part of the fifth edition of the WHO “blue books”.

Read more in a summary in Cancer Discovery


Opioid fears a barrier to cancer analgesia adherence

Non-adherence with pain management in head and neck cancer is common and driven in part by fears of addiction, Australian researchers say.

A small retrospective chart review over two years in a regional cancer clinic found 70% of patients were non adherent with prescribed analgesia despite reporting severe pain.

“Three main reasons for non-adherence with pain management were reported: Fear of addiction and resistance to using opioids, administration or dose confusion, and difficulty organising medications (such as filling scripts).”

The researchers said there may be unique characteristics among regional patients that warrant attention.

Clinicians require early identification strategies to identify and intervene early to address opioid stigma and fear. In rural settings, the unique contextual factors which influence this fear and stigma need consideration.”

Read more in Collegian


Better than a doctor? IBM abandons ‘Watson’ oncology AI program

A decade after declaring its AI system was better at diagnosing cancer than a doctor, IBM has dumped its troubled Watson supercomputer business.

Developed in conjunction with the Memorial Sloan Kettering Cancer Center, the Watson for Oncology system was once claimed to have a 90% accuracy rate for diagnosing lung cancer compared to 50% for human doctors. Its developers said human doctors would need to spend 160 hours a week reading clinical studies to keep up with the latest developments in oncology trials whereas Watson’s supercomputing power allowed it to quickly absorb and apply this knowledge in clinical situations.

However in real world use the Watson Health AI programs had operational problems, needed extensive human input and only worked well for uncommon cancers.

IBM has since shut down the oncology program and this week sold off the remaining healthcare data and analytics parts of Watson Health to investment firm Francisco Partners.

Already a member?

Login to keep reading.

OR
Email me a login link
logo

© 2022 the limbic