Tag

Tagged: image recognition

Sponsored
  • People and doctors often miss early warning signs of cancer
  • Nearly 50% of all cancers are diagnosed late when they have already spread
  • Each year cancer kills 8m people worldwide and cost billions
  • 40% of cancer deaths could be prevented by early detection
  • Traditional tissue biopsies used to diagnose cancer are invasive, slow, costly and often yield insufficient tissue
  • New non-invasive tests are being devised to detect cancer early
  • Such tests are positioned to significantly reduce the vast and growing global burden of cancer
  • But before these tests enter clinics, they need to overcome a number of challenges
 

A paradigm shift in cancer diagnosis


How close are we to developing a simple, cheap, rapid and exquisitely sensitive non-invasive test to diagnose cancer in healthy-looking people?

Recently, attention has been drawn to a breathalyser test for cancer diagnosis, which is just starting a significant 2-year clinical study in the UK. In 2018, a “liquid biopsy” was popularly heralded as “the holy grail” of cancer diagnosis, only quickly to be quashed by medical experts who warned that this conclusion was “premature” and “misleading”. Further, image recognition is increasingly being used as a technique to detect cancer. Given the extent and depth of these endeavours it seems reasonable to assume that, within the next decade, gold-standard solid tumour biopsies for detecting cancer will be replaced by non-invasive diagnostic techniques.

 
In this Commentary
 
In this Commentary we describe evolving innovative techniques to detect cancer early, which include a breathalyser, a liquid biopsy and an image recognition test. But first we: (i) briefly describe the epidemiology of cancer, (ii) explain the extent, implications and some of the causes of late diagnosis, which is driving the development of these new non-invasive detection techniques, (iii) describe how ‘personalized’ medicine, predicated upon the molecular signatures of cancer tumours, has become routine clinical practice and demand more efficacious techniques to understand the complexities of cancer.
 
Cancer snapshot
 
Cancer is among the leading causes of death worldwide. In 2012, there were 14.1m new cases and 8.2m cancer-related deaths worldwide. 57% of these new cancer cases occurred in less developed regions of the world, which include Central America, parts of Africa and Asia. 65% of cancer deaths occurred in these regions. The number of new cancer cases per year is expected to rise to 23.6m by 2030. It is estimated that over 40% of cancer cases are preventable. In the UK there are more than 360,000 new cancer cases and over 166,000 cancer deaths every year. Since the early 1990s, incidence rates for all cancers combined in England have increased by 13% each year. Annual NHS costs for cancer services are over £5bn, but the cost to British society - including costs for loss of productivity - is over £18bn. In the US, over 1.7m new cases of cancer were diagnosed in 2018 and some 0.61m people died from the disease. It is estimated that in the US the annual national expenditure on cancer is some US$150bn. Early diagnosis and cancer prevention would significantly reduce  cancer morbidity and mortality and achieve large cost savings for healthcare systems.
 
The challenge of late cancer diagnoses

The significance of developing a simple non-invasive test to diagnose cancer early cannot be over-emphasised. For a number of reasons, almost half of people who get cancer are diagnosed late, which makes treatment less likely to succeed, reduces chances of survival and significantly increases the cost of care. For instance, in the UK about 25% of all cancer cases only are diagnosed following presentation in A&E. The vast majority of these cases are already at a late stage, when treatment options are limited, and survival is poorer. Further, a substantial percentage of people neither avail themselves  of cancer screening nor present themselves to primary care physicians with early symptoms. A good example of this is cervical cancer screening in the UK, which is offered every three years to all women aged between 25 and 64. Despite the test only taking a few minutes, each year over 1.3m women choose not to attend, and non-attendance is the biggest risk factor to developing cervical cancer. Each year, some 220,000 women in the UK are diagnosed with cervical abnormalities and over 800 women die from the disease.
 
Implications of inefficient healthcare systems
 
Late diagnosis not only occurs for non-compliance. Some cancers are asymptomatic while others have general non-specific symptoms and are often mistaken for lesser ailments. Further, inefficiencies in healthcare systems can lead to late diagnosis and increased cancer morbidity and mortality. For example, in February 2019 the UK’s National Audit Office (NAO) published an “Investigation into the management of health screening”, which concluded that none of the key screening programs in England - for bowel, breast or cervical cancer - met their targets because of management and IT failures.  As a consequence, about 3m women across England have not had a cervical cancer test for at least three-and-a-half years. In 2018, more than 150,000 cervical screening samples piled-up in laboratories due to outdated IT systems, staff shortages and changes in testing procedures. Faulty IT systems also are reported to have resulted in 5,000 women not being invited for breast screening, which in England is currently offered once every three years to women aged 50 to 70. According to the NAO report, in 2017 450,000 women missed a final breast cancer screening test because of a system failure, which is believed to have been responsible for some 270 deaths.
 
Molecular biology challenges to gold standard solid tissue biopsies
 
In the past decade, ‘personalized’ medicine predicated upon the molecular signatures of cancer tumours has become routine clinical practice. The identification on tumour tissue of predictive biomarkers of response to personalized targeted therapies is now considered optimal patient care. Notwithstanding, such treatment faces a number of biological and technological challenges associated with traditional solid tumour biopsies' access to tumours and the heterogeneity of tumours.
 
While some cancer tumours are easily accessed, others have limited accessibility because they are either deep in the body or embedded in critical organs. This makes obtaining a comprehensive “picture” of such tumours challenging and may increase clinical complications. Further, tissue samples from different regions of the same tumour may differ and tissue specimens from primary and metastasized tumours can also differ. In addition, studies have shown the dynamic changes of tumour features over time and the emergence of therapy-resistance. Thus, inter- and intra-tumour heterogeneity pose a pivotal challenge to guide clinical decision-making in cancer therapy as traditional biopsies may be unable to capture a complete genomic landscape of a patient’s tumour. 
 
A non-invasive test, such as sampling blood, urine, salvia and breath can provide the same genetic information as a solid tissue biopsy and has certain added advantages, which include: (i) they are a source of fresh tumour-derived material, unhampered by preservatives and (ii) they provide an alternative sample type in routine clinical practice when tumour sampling is unavailable, inappropriate or difficult to obtain.
Breath test to diagnose cancer

Because of the challenges associated with traditional biopsies, clinical attention is turning to non-invasive tests and recently to a breath test, which promises to be able to diagnose cancer early. A study to detect cancer through breath, which was carried out by researchers from Imperial College London and the Karolinska Institutet in Sweden and presented at the 2017 European Cancer Congress (ECC) in Amsterdam, Holland was promising but inconclusive.

You might also like:

World’s first blood tests that detect and locate cancer


 
The study aimed to test whether a “chemical signature”, composed of five substances, which seemed to typify cancer could be the basis for a diagnostic test for the disease. Breath samples were tested of some 335 patients attending leading London hospitals. Of these, 163 had been diagnosed with oesophageal or stomach cancer and 172 presented with upper gastrointestinal symptoms, but without any evidence of cancer after an endoscopy.

Findings suggested that four of the five chemical substances were expressed differently in the breath samples from those diagnosed with cancer, compared to those where no cancer had been found. The breath test was able to correctly indicate cancer in around 80% of patients who had cancer (sensitivity), and able to correctly exclude cancer in around 80% of cases, which did not have cancer, (specificity). Although the findings were promising, researchers concluded that, "The study shows the potential of breath analysis in non-invasive diagnosis of oesophageal cancer. The potential benefits of this technology to patients may be early diagnosis and improved chance of survival. If placed as an endoscopy triage test, the benefits to healthcare systems may include cost-saving through reducing the number of negative endoscopies. However, these findings must be further validated in an un-enriched larger population of patients undergoing diagnostic endoscopy and in false negative patients the value of repeat testing should be established".

 
Expanded clinical study for breath biopsy

Following these promising conclusions a large two-year clinical trial of a breath test, called the Breath Biopsy, supported by Cancer Research UK was started in January 2019 at Addenbrooke’s Hospital in Cambridge, UK, and aims to detect whether exhaled airborne molecules called volatile organic compounds (VOCs) can be useful in detecting cancer. The study expects to recruit 1,500 participants including healthy people to act as a control group. Scientists hope the study will lead to a simpler, cheaper method of spotting cancers at an early stage when they are more likely to respond to treatment. Study participants will be asked to breathe into a device called the Breath Biopsy, which has been developed by Owlstone Medical, a private company founded in 2003 and based in Cambridge. 
 
Breath biopsy to target two challenging cancers
 
In the first instance, only patients with oesophageal and stomach cancers will be invited to try the Owlstone breath biopsy.  Both of these cancers are aggressive and tend to be diagnosed late because in the early stages they either cause no symptoms - in the case of oesophageal cancer - or symptoms that are vague and easy to mistake for other less serious conditions - in the case of stomach cancer. Currently, oesophageal and stomach cancers are diagnosed using endoscopy, which involves a camera attached to a flexible tube being passed down the throat. The procedure is invasive, it risks complications and is expensive. If the breath test is successful with these two cancers, it will be expanded to include patients with prostate, kidney, bladder, liver and pancreatic cancers.
 

Oesophageal and stomach cancers
 
Oesophageal cancer is the 7th most commonly occurring cancer in men and the 13th most commonly occurring cancer in women. In 2018, there were over 0.5m new cases diagnosed globally. The 5-year survival rate for patients with oesophageal cancer is less than 20%. Each year, there are around 9,000 new cases diagnosed in the UK and around 7,900 oesophageal cancer deaths. In the US, it is estimated that there were 17,290 new cases of the disease and 15,850 deaths in 2018
 

Stomach cancer is the 4th most commonly occurring cancer in men globally and the 7th most commonly occurring cancer in women. The disease represents the 3rd cause of cancer death in the world with about 723,000 deaths each year, which accounts for 8.8% of all cancer deaths. In 2018, there were over 1m new cases of stomach cancer worldwide. The five-year survival rate for the disease is about 30% and the 10-year survival rate 15%. According to the American Cancer Society's estimates, over 27,000 patients are expected to be diagnosed with stomach cancer in the US in 2019, of whom some 11,000 are expected to die. In the UK, there are around 7,000 new stomach cancer cases every year and around 4,500 stomach cancer deaths.

 
Chemical signature

According to the lead investigator of the breath biopsy clinical study, Rebecca Fitzgerald, professor of Cancer Prevention at Cambridge University and Consultant in Gastroenterology and General Medicine at Addenbrooke's Hospital, “We urgently need to develop new tools, like this breath test, which could help to detect and diagnose cancer earlier, giving patients the best chance of surviving their disease. Through this clinical trial we hope to find signatures in breath needed to detect cancers earlier. It’s the crucial next step in developing this technology.”
 
Liquid biopsies

Following the presentation of research findings of liquid biopsy clinical studies at the 2018 annual conference of the American Society of Clinical Oncology (ASCO) there were press reports suggesting that the new test was the holy grail” of cancer diagnosis. This was quickly quashed by medical experts who described the press claims as “premature” and “misleadingly”.

A “liquid biopsy“ has the potential to detect and classify mutations from minute fragments of circulating tumour in a blood sample and entails assessing circulating tumour cells (CTCs) and cell-free DNA (cfDNA) and its subsets of circulating tumour DNA (ctDNA) and cell-free RNA (cfRNA). Liquid biopsies are considered to provide significantly superior biomarkers than the traditional cancer biomarkers such as the prostate specific antigen (PSA) and cancer antigen 125 (CA125) tests, which have been used for decades to support the diagnosis and management of cancer. With the exception of the PSA test, which is used as a screening test for prostate cancer, none of the traditional cancer tests are recommended for population screening because their sensitivity and specificity are not accurate enough.

 
Liquid biopsies effective only after diagnosis
 
While promising, liquid biopsies represent an emerging technology, which has been shown to be effective in personalizing therapy after diagnosis but has yet to demonstrate its clinical utility against the current gold standard tissue biopsies for confirming a cancer diagnosis. There is a relative dearth of evidence on the capabilities of liquid biopsies for detecting cancer early. Expert consensus suggests that liquid biopsies have significant limitations and the tests are not sufficiently developed for widespread use. Liquid biopsies are neither as good nor better than existing screening methods and are not ready for meaningful clinical application because their accuracy, reliability, and reproducibility are still unknown.
 
US biotech start-up conducting large clinical studies of liquid biopsies
 
Notwithstanding, the development of liquid biopsies continue at a pace. Not least the R&D being undertaken by GRAIL Inc., a private US biotech company, spun out in 2015 from San Diego-based Illumina, the world’s largest gene sequencing company. GRAIL is currently valued at US$2.5bn and since its inception has raised US$1.5bn. The company has started two large long-term clinical studies aimed at developing a liquid biopsy for early cancer detection.
 
Early test results suggest that it is not money holding these liquid biopsies back, but basic biology. To evaluate potential blood screens, thousands of patients will have to get tested - and then researchers will have to wait for some of them to actually get cancer, which is the only way to determine not only the predictive power of the tests, but also whether they lead to improved patient outcomes.
 
Image recognition and medical diagnosis
 
Image recognition is another technology being used to develop non-invasive cancer diagnostic tests. Examples include Google’s Lymph Node Assistant (LYNA), which claims to be better than doctors at spotting late-stage breast cancer. LYNA can detect secondary cancer cells in medical scans with 99% accuracy. Secondary cancer cells are responsible for spreading cancer and detecting them is time-consuming and challenging for pathologists.

A study published in the August 2018 edition of Nature Mind reports findings on the first phase of a study undertaken by Moorfields Eye Hospital and Google’s DeepMindwhich enables computers to analyse high-resolution 3D scans of the back of the eye to detect more than 50 eye conditions. 

A study published in the October 2017 edition of the journal Frontiers in Psychology, report findings of research conducted by scientists from  Macquarie University in Sydney, Australia, which suggests facial shape analysis can correctly detect markers of physiological health in individuals of different ethnicities.

Shanghai based Yitu Technology and Beijing-based Infervision are among start-ups racing to improve medical imaging analysis by using the same technology that powers facial recognition and autonomous driving. These examples, and others, are indicative of an intensifying competition between the US and China to dominate the life sciences, which is a significant growth industry of the future. In a forthcoming Commentary we shall describe this competition in more detail and explain the comparative advantages of the two nations.

 
Takeaway

It seems reasonable to suggest that over the next decade the gold standard solid tissue biopsy for diagnosing cancer will be replaced with cheap, rapid, non-invasive diagnostic tests, which are able to detect cancer early and thereby make a significant dent in the vast and escalating global burden of the disease.
view in full page