There is growing interest in tumour markers as aids for the diagnosis, staging and management of cancer. Some are expected to succeed, after several years of evaluation to trials and eventual clinical use. A large number, however, are not likely to make it beyond development. On their part, physicians need to be aware of both opportunities and limitations in the clinical use of tumour markers.
Tumour markers are substances (antigens, proteins, enzymes or hormones) which indicate the presence of cancer or provide information about its likely course of development. They are present in cancerous tissue as well as in the bodily fluids of cancer patients.
Range of applications
Tumour markers have shown their potential for several applications. These range from the differential diagnosis of benign and malignant conditions to prognostic assessments, postoperative surveillance, the prediction of drug response or resistance, and the monitoring of therapy in
advanced disease.
The key advantage of tumour markers in the above applications is convenience. Inexpensive automated assays allow for fast processing of samples.
The case for tumour markers
The best known tumour markers include Her2/neu for breast cancer, which has an established economic case. Her-2/neu is a target for trastuzumab, whose use as an adjuvant has been shown to decrease cancer recurrence rates by 50%. However, up to one in 20 trastuzumab recipients develop cardiac dysfunction. Given that the cost of one year of therapy is close to 100,000 Euros, the need for accurately and precisely assaying every tissue sample is evidently strong.
Work in progress
Tumour markers are, however, still a work in progress and expected to remain so. In the US, the National Cancer Institute (NCI) states that “more than 20 tumour markers are currently in use.” It however lists over 30. The European Group on TumoUr Markers (EGTM) has a list of 16.
Despite the number of tumour markers in development, only ‘traditional’ markers are used in diagnosis, prognosis and monitoring. For example, at least six urine tumour marker kits are approved by the US Food and Drug Administration for bladder cancer. However, none are backed by data from clinical trials that increased survival time, improved quality of life or decreased cost of treatment.
Appropriate use, caution urged
Many experts urge caution with respect to tumour markers. Inappropriate use, according to an article in the ‘British Medical Journal’, can cause patients unnecessary anxiety and distress, and may also delay correct diagnosis and treatment. The authors cite one hospital audit which found “that only about 10% of requests for tumour markers were appropriate.”
The European Group on Tumor Markers attributes part of this problem to the growing availability of automated immunoassays. This makes tumour marker tests available in routine rather than specialist laboratories. “Results are consequently more readily available to non-specialist clinicians, who may be less familiar with their interpretation.”
Challenges of sensitivity and specificity
Only some markers, known as tumour-specific markers, are produced exclusively by a particular tumour As a result, most tumours cannot be detected by a single test, and tests for multiple markers are often required.
Tests are therefore often accompanied by the risk of both false positives and false negatives. The Cancer Information & Support Network (CISN) sums up the picture: False positives may occur because most tumour markers “can be made by normal cells, as well as cancer cells,” and markers “can be associated with noncancerous conditions.” On the other hand, the reason for false negatives is that “tumour markers are not always present in early stage cancers” and because “people with cancer may never have elevated tumour markers.”
For example, the level of CA-125, a marker for ovarian cancer, is also elevated in a variety of non-malignant disorders such as cirrhosis, pancreatitis, endometriosis, and pelvic inflammatory disease. In addition, medications appear to alter the results of a varied range of tests. So too do pregnancy, menstruation, cigarette smoking and various benign disorders.
Biopsy remains only definitive way for diagnosis
The above lack of sensitivity and specificity has been a major limitation facing the use of tumour markers in clinical practice. As with imaging, the use of tumour markers has been limited to supporting the diagnostic process, and the gold standard for diagnosis still remains a biopsy.
Although difficult to access areas such as the brain are likely to result in more use of tumour markers, a biopsy remains “the only definitive way” for diagnosis of a tumour” even in the brain.
NACB Guidelines
In 2008, the National Academy of Clinical Biochemistry (NACB) in the US released updated Laboratory Medicine Practice Guidelines for the use of tumour markers.
The guidelines made cross-referrals to efforts by numerous professional and regulatory best-practices bodies, including the American Society of Clinical Oncology (ASCO) and the National Comprehensive Cancer Network (NCCN), Britain’s National Institute for Health and Clinical Excellence (NICE), the European Group on Tumor Markers (EGTM), the International Federation of Gynecology and Obstetrics (FIGO) and the Gynecologic Cancer Intergroup (GCIG).
The NACB guidelines cover five cancer sites: testicular, prostate, colorectal, breast, and ovarian.
Testicular cancer
For testicular cancer, α-fetoprotein (AFP), human chorionic gonadotropin and lactate dehydrogenase are recommended for diagnosis, staging, prognosis determination, recurrence detection and the monitoring of therapy. AFP is also recommended for the differential diagnosis of tumours
Prostate cancer
Prostate-specific antigen (PSA) is considered to be potentially useful for detecting prostate cancer recurrence and monitoring therapy. Free PSA is considered useful for distinguishing malignant from benign prostatic disease.
Colorectal cancer
In colorectal cancer, carcinoembryonic antigen is recommended (with some caveats) for prognosis determination, post-operative surveillance, and therapy monitoring in advanced disease. Fecal occult blood testing is considered useful for screening asymptomatic adults who are older than 50 years.
Breast cancer
For breast cancer, estrogen and progesterone receptors predict response to hormone therapy, human epidermal growth factor receptor-2 predicts response to trastuzumab, while urokinase plasminogen activator/ plasminogen activator inhibitor 1 is used for determining prognosis in lymph node-negative patients. CA15-3/BR27–29 or carcinoembryonic antigen can be used for therapy monitoring in advanced disease.
Ovarian cancer
CA125 is recommended (with transvaginal ultrasound) for early detection of ovarian cancer in women at high risk for this disease. CA125 is also recommended for differential diagnosis of suspicious pelvic masses in post-menopausal women, as well as for detection of recurrence, monitoring of therapy, and determination of prognosis in women with ovarian cancer.
Future research to target higher specificity and sensitivity
In the future, research is expected to focus on finding markers which are specific of one pathology, have higher sensitivity with a low cut-off and deliver results which correlate to tumour mass and growth potential. Ideal candidates would also have a short life duration to permit efficient follow-up; in other words, their presence should decrease during treatment and increase before a relapse.
The promise and challenges of screening
Given that tumour markers can aid in assessing the response to cancer treatment and making prognoses, many public health professionals have hoped they might also be used for screening tests which would detect cancer before the presence of symptoms.
Indeed, many tests have both screening and diagnostic uses, with only the context of use determining whether the test is one or the other. “A screening test is done on asymptomatic individuals who receive the test principally because they are of the age or sex at risk for the cancer. A diagnostic test is done on an individual because of clinical suspicion of disease.”
However, no tumour marker identified to date is sufficiently sensitive or specific to be used on its own for screening, demonstrating a survival benefit in randomized controlled trials in the general population.
One of the best known examples is the prostate-specific antigen (PSA) test. Although it is now accepted that most men with elevated PSA levels do not have prostate cancer, the implications of this remain mired in controversy and also illustrate the kind of limitations which other tumour biomarkers may face in the future for use in screening.
PSA screening has been the subject of two large randomized controlled trials in the US and Europe in the 2000s. However, as the Mayo Clinic notes, in spite of the size of the trials, there were “no clear conclusions.” This is because their diversity of methodology allows for significant flexibility in interpretation. As a result, “the decision of whether to screen or not screen – using PSA testing or other means or both – is a decision best made between physicians and their individual patients.”
The two trials were PLCO (Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial) conducted by the National Cancer Institute in the US, and the European Randomized Study of Screening for Prostate Cancer (ERSPC), billed as “the largest randomized trial of screening for prostate cancer” with 162,388 subjects.
In 2011, a US government task-force concluded that healthy men should not be screened for prostate cancer. The finding, which “drastically changed the standard of care for middle-age American men who had grown accustomed to annual screenings,” was largely based on 10 years data from the two studies, which found risk of over-diagnosis and over-treatment.
Problems began when follow-on data from ERSPC two years later showed screening was associated “with a 21% reduction in risk of prostate cancer mortality.” However, this was accompanied by a still-sizeable risk of over-diagnosis and over-treatment. As a result, the authors said that “population-based screening could not yet be recommended.
In April 2015, an article in ‘The Lancet’ re-confirmed “a substantial 21% reduction” from PSA screening. However, due to access restrictions to the ERSPC trial data, the authors called this figure into serious question.
The controversy is unlikely to go away for some time. An Op Ed in The New York Times called the PSA test “hardly more effective than a coin toss.” Although the date of publication was 2010, the author of the commentary was Dr. Richard Ablin, who discovered PSA in 1970.
Such challenges are also likely to accompany screening for other conditions. For instance, data from the PLCO trial show that screening for CA-125 (recommended by the National Academy of Clinical Biochemistry for women with ovarian cancer, along with transvaginal ultrasound), does not reduce ovarian cancer mortality. Instead, false-positive screening test results have been associated with complications.
Electronic health records and the lab
, /in Featured Articles /by 3wmediaTwo opposing agendas confront clinical labs in terms of electronic health records (EHRs): privacy/security on the one side, and interoperability, on the other. The former involves an inward push for isolation, while the latter tends to pull technology in the other direction.
There also is a major financial challenge. While healthcare providers have been given a host of incentives to adopt EHRs (especially in the US), labs have been pretty much left out on their own.
EHRs and lab systems populate different worlds
Clearly, lab-compatible EHR systems which meet both (privacy and interoperability) criteria promise the quickest returns. EHR developers have however shown little enthusiasm, until recently, to incorporate clinical lab requirements as a sufficient driver, while laboratory system vendors have tended to ignore EHRs or postpone taking them into account until EHR development has matured sufficiently.
US EHR adoption drives lab applications
In the US, this limbo is being shaken up by healthcare providers, who are compelling vendors to take account of their need for EHR-friendly clinical lab systems.
At end 2012, the US Centers for Disease Control and Prevention (CDC) released a survey which found 72 percent of office-based physicians using EHR systems, up from 48 percent in 2009 and 18 percent in 2001.
The reason for the dramatic increase in EHR adoption lies in the Meaningful Use requirements of the 2009 Health Information Technology for Economic and Clinical Health Act, also known as the HITECH Act. The Act provides billions of dollars in incentive payments through the Medicare and Medicaid programmes to increase physician adoption of EHR systems.
Clinical labs are now being lifted by the rising tide of EHR adoption. According to the US Office of the National Coordinator for Health Information Technology (ONC), the “availability of structured lab results within the EHR contributes to office efficiencies while also assisting providers in the ability to make real time decisions about the patient’s care.”
The ONC explicitly specifies the threshold for EHR-friendly clinical lab practices in Stage 1 – of over 40 percent of all lab test results ordered by a provider and incorporated in certified EHR technology as structured data.
Stage 2 Meaningful Use requirements, finalised in August 2012, increase the clinical lab results threshold to 50 percent. The ONC has subsequently announced plans to assess health information exchange (HIE) in clinical laboratories.
Labs left to own resources
While healthcare providers have the financial incentives of the HITECH Act, clinical labs have been left to their own resources to set up interfaces from their laboratory information systems (LIS) to providers.
Compounding this has been inconsistencies in the way different EHR systems generate lab test orders.
However, the alternative has been stark – to be left out of referrals from tests.
EHR systems remain heterogeneous
The US EHR landscape is however hardly uniform. As of September 2013, there were 3,652 non-enterprise certified ambulatory EHR software systems, almost half of which were classified as “complete” to qualify for Meaningful Use Stage 1 or Stage 2.
In spite of efforts to set standards for semantic interoperability of healthcare data, standards so far are only syntactic (based on HL7 and XML).
The alternative, to develop a common US-wide EHR system, has been accepted as being technically insurmountable – due to hurdles in specifying, developing, testing and deploying standardized tools, common architectures and vocabularies, within secure, real-time and scalable networks, and doing all this within the fast-changing world of information and communications technologies.
For proponents of a decentralized approach to EHR technology, in the US in particular, the sharp increase in offtake of EHR systems has shown that it has delivered – as far as healthcare IT objectives are concerned.
EHR faces teething problems
Still, teething troubles for EHRs also clearly remain.
In early September 2013, one of the leading EHR systems, from EPIC, crashed across seven major healthcare facilities of Sutter Health, a nearly 100 year-old healthcare provider in California. Some suspect the role of a routine upgrade a few days earlier in the EHR system, which was launched by Sutter at a cost of $1.2 billion in 2004, but has so far reached only a halfway mark.
EHR challenges for labs remain to be resolved
Such issues with the evolution of maturity of EHRs pose especially major problems for labs, who (as mentioned) have to develop and fund interfaces between their LISs and the EHRs of their client physicians but are also forced to cope with the lack of uniform EHR standards.
Some vendors have nevertheless sought to fill the gap.
A leading example is HDD Access, a joint initiative by the US Department of Defense, the Department of Veterans Affairs and 3M Health Information Systems to create a public use version of 3M’s Healthcare Data Dictionary (HDD). HDD Access consists of a relational database and Application Programming Interface (API) runtime services to which other applications can interface. The terminology is organized as a controlled medical vocabulary – a comprehensive set of clinical and other concepts used in healthcare.
HDD Access offers specific benefits for integrating LIS and EHR platforms. Independent of source system, it can track local fields and translate them into laboratory concepts. Nevertheless, HDD Access warns that it is “not a standard terminology and is not a replacement for standard terminologies.
In effect, in the US, clinical labs are likely to continue to face a host of technical challenges with respect to EHRs in the years to come.
EHR Big Bang fizzles in Europe
Unlike the US, Europe made a massive effort in 2004 to devise common semantic standards for EHR interoperability as part of its Single eHealth Area. The EU’s EHR objectives sought to integrate all patient information – from primary to tertiary settings, and include emergency and in-patient care. Also on the radar were ambitious plans to connect pharmacies as well as the web of disparate billing/reimbursement procedures, and do so across Europe.
In mid–2008, the EU Commission set 2015 as the target year for EHR interoperability, to ensure that key EHR datasets could cross European borders, and do so in conformity with medical rules and other relevant legal frameworks.
In January 2011, however, these ambitions were put on the backburner, after an official report criticized the effort as being both impractical and ‘grandiose’. The report found that a pan-EU EHR system would neither be technically feasible, cost-effective or even medically justified, and instead urged more emphasis on decentralized efforts – in other words, just like the US.
Technical challenges aside, massive differences in physician and medical cultures across Europe played a major role in derailing efforts toward a common EHR. Or, as EuroRec, an umbrella organization tasked with pan-EU EHR implementation, states: it was “widely recognized that social and organizational aspects are as likely to ruin an implementation process as technical factors are.”
European focus shifts to national efforts
The EHR focus in Europe has now totally shifted to national efforts. A new eHealth Governance Initiative (eHGI) encourages cooperation “between Member States” and “between national authorities and standardization bodies”, and seeks to “enable the recommendation of standards and (harmonized) profiles based on selected use cases.” On the technical side, compared to the Big Bang efforts of the Single eHealth Area, it also aims to “link and harmonize coding systems” and “facilitate access to existing standards and medical vocabularies.”
The second area for Europe’s EHR focus is a minimalistic intra-EU/regional approach embodied in a project called epSOS, which dates back to 2008, but was (temporarily) eclipsed by the ambitions of the Single eHealth Area. epSOS, which went live in April 2012, has the modest goal of connecting 20 EU nations (and 3 non-EU members) to a secure database, and sharing only Patient Summaries and ePrescription records via IHE X* profiles. Its target consists of Europeans holidaying overseas.
Today, EHR adoption varies considerably in Europe. The Nordic countries have been using the technology for over a decade and are fairly advanced as a result in EHR implementation.
However, adoption in France, Germany, Spain and the UK is ‘on course’ with the US.
Shift from Single eHealth Area encourages new EHR-directed lab applications
The shift away from forcing through a Single eHealth Area has also opened the way for innovative working approaches aimed at clinical labs. One good example of this is Valle de los Pedroches Hospital at Cordoba, Spain, which has designed and implemented a unified lab test request module for the Andalusian regional EHR.
In spite of some outstanding issues (such as rigidity in error solving, and the need to adapt to a new nomenclature), implementation of the laboratory module in the EHR improved the analytical process, with better patient safety and less programming or container errors and shorter response times. Clinical professionals gave a rating of 7.8 out of 10, positively highlighting the speed at which results are delivered and their integration in the EHR.
Such efforts are likely to grow with time.
Eco Series: The next generation freezers
, /in Featured Articles /by 3wmediaThink Coag – Think Tcoag
, /in Featured Articles /by 3wmediaStat Profile Prime
, /in Featured Articles /by 3wmediaCXCL13 and anti-Borrelia
, /in Featured Articles /by 3wmediaIMMUVIEW
, /in Featured Articles /by 3wmediaUriSed mini
, /in Featured Articles /by 3wmediaTime matters, results count
, /in Featured Articles /by 3wmediaTumour markers: why not VOCs?
, /in Featured Articles /by 3wmediaSince early diagnosis of cancer is crucial in reducing mortality, much research is directed towards finding minimally invasive methods of detecting early tumour growth. In 1971 the Nobel Prize winner Linus Pauling froze his exhaled breath to analyse the volatile organic compounds (VOCs) it contained by gas chromatography. Forty-six years later, although many studies have suggested that some of the over two hundred VOCs found in human breath and other body specimens such as urine and stool could be diagnostic for different cancers, routine VOC analysis has yet to reach clinical laboratories.
A fairly recent study involving over 300 subjects demonstrated that appropriately trained dogs could identify cases of colon cancer from breath or stool samples with 95% accuracy, and there is a plethora of anecdotal evidence suggesting that canines can identify breast, ovarian, bladder, skin and lung cancers by smelling appropriate samples. Last month an article in the journal of Urology reported that two female dogs had correctly identified prostate cancer in well over 90% of cases by sniffing urine samples from 900 men, 360 of whom had the disease. While packs of Labradors roaming clinic corridors would be neither practical nor affordable, this is surely evidence that signature VOCs for different tumours do exist, and that their detection would be an excellent non-invasive approach for early cancer diagnosis if suitable methods could be developed.
So what are the problems? Firstly much of the research on cancer in recent years has understandably focused on changes in DNA and proteins in tumour cells. In addition relevant VOCs are found in such low concentrations (parts per billion/trillion) that research laboratories with expensive instruments such as GC-MS, proton transfer reaction-MS and selected ion flow tube-MS have so far been necessary to extract and analyse them. A study published in last month’s Gut reported the analysis of VOC patterns in the exhaled breath of 484 subjects, 99 of whom had gastric cancer and 325 of whom had pre-cancerous conditions, by two methods: GC-MS and via a nanoarray sensor. The results were encouraging, indicating that an eight VOC signature had a specificity of 98%; sensitivity was 73%. And a small sorbent trap for breath collection has now been developed that captures more than two hundred VOCs for subsequent GC-MS analysis to find the elucidated signatures specific for various diseases including breast and lung cancer. Clinical trials are underway; the goal is a point-of-care system for early diagnosis. After nearly half a century there may be light at the end of the tunnel!
The promise and the challenges
, /in Featured Articles /by 3wmediaThere is growing interest in tumour markers as aids for the diagnosis, staging and management of cancer. Some are expected to succeed, after several years of evaluation to trials and eventual clinical use. A large number, however, are not likely to make it beyond development. On their part, physicians need to be aware of both opportunities and limitations in the clinical use of tumour markers.
Tumour markers are substances (antigens, proteins, enzymes or hormones) which indicate the presence of cancer or provide information about its likely course of development. They are present in cancerous tissue as well as in the bodily fluids of cancer patients.
Range of applications
Tumour markers have shown their potential for several applications. These range from the differential diagnosis of benign and malignant conditions to prognostic assessments, postoperative surveillance, the prediction of drug response or resistance, and the monitoring of therapy in
advanced disease.
The key advantage of tumour markers in the above applications is convenience. Inexpensive automated assays allow for fast processing of samples.
The case for tumour markers
The best known tumour markers include Her2/neu for breast cancer, which has an established economic case. Her-2/neu is a target for trastuzumab, whose use as an adjuvant has been shown to decrease cancer recurrence rates by 50%. However, up to one in 20 trastuzumab recipients develop cardiac dysfunction. Given that the cost of one year of therapy is close to 100,000 Euros, the need for accurately and precisely assaying every tissue sample is evidently strong.
Work in progress
Tumour markers are, however, still a work in progress and expected to remain so. In the US, the National Cancer Institute (NCI) states that “more than 20 tumour markers are currently in use.” It however lists over 30. The European Group on TumoUr Markers (EGTM) has a list of 16.
Despite the number of tumour markers in development, only ‘traditional’ markers are used in diagnosis, prognosis and monitoring. For example, at least six urine tumour marker kits are approved by the US Food and Drug Administration for bladder cancer. However, none are backed by data from clinical trials that increased survival time, improved quality of life or decreased cost of treatment.
Appropriate use, caution urged
Many experts urge caution with respect to tumour markers. Inappropriate use, according to an article in the ‘British Medical Journal’, can cause patients unnecessary anxiety and distress, and may also delay correct diagnosis and treatment. The authors cite one hospital audit which found “that only about 10% of requests for tumour markers were appropriate.”
The European Group on Tumor Markers attributes part of this problem to the growing availability of automated immunoassays. This makes tumour marker tests available in routine rather than specialist laboratories. “Results are consequently more readily available to non-specialist clinicians, who may be less familiar with their interpretation.”
Challenges of sensitivity and specificity
Only some markers, known as tumour-specific markers, are produced exclusively by a particular tumour As a result, most tumours cannot be detected by a single test, and tests for multiple markers are often required.
Tests are therefore often accompanied by the risk of both false positives and false negatives. The Cancer Information & Support Network (CISN) sums up the picture: False positives may occur because most tumour markers “can be made by normal cells, as well as cancer cells,” and markers “can be associated with noncancerous conditions.” On the other hand, the reason for false negatives is that “tumour markers are not always present in early stage cancers” and because “people with cancer may never have elevated tumour markers.”
For example, the level of CA-125, a marker for ovarian cancer, is also elevated in a variety of non-malignant disorders such as cirrhosis, pancreatitis, endometriosis, and pelvic inflammatory disease. In addition, medications appear to alter the results of a varied range of tests. So too do pregnancy, menstruation, cigarette smoking and various benign disorders.
Biopsy remains only definitive way for diagnosis
The above lack of sensitivity and specificity has been a major limitation facing the use of tumour markers in clinical practice. As with imaging, the use of tumour markers has been limited to supporting the diagnostic process, and the gold standard for diagnosis still remains a biopsy.
Although difficult to access areas such as the brain are likely to result in more use of tumour markers, a biopsy remains “the only definitive way” for diagnosis of a tumour” even in the brain.
NACB Guidelines
In 2008, the National Academy of Clinical Biochemistry (NACB) in the US released updated Laboratory Medicine Practice Guidelines for the use of tumour markers.
The guidelines made cross-referrals to efforts by numerous professional and regulatory best-practices bodies, including the American Society of Clinical Oncology (ASCO) and the National Comprehensive Cancer Network (NCCN), Britain’s National Institute for Health and Clinical Excellence (NICE), the European Group on Tumor Markers (EGTM), the International Federation of Gynecology and Obstetrics (FIGO) and the Gynecologic Cancer Intergroup (GCIG).
The NACB guidelines cover five cancer sites: testicular, prostate, colorectal, breast, and ovarian.
Testicular cancer
For testicular cancer, α-fetoprotein (AFP), human chorionic gonadotropin and lactate dehydrogenase are recommended for diagnosis, staging, prognosis determination, recurrence detection and the monitoring of therapy. AFP is also recommended for the differential diagnosis of tumours
Prostate cancer
Prostate-specific antigen (PSA) is considered to be potentially useful for detecting prostate cancer recurrence and monitoring therapy. Free PSA is considered useful for distinguishing malignant from benign prostatic disease.
Colorectal cancer
In colorectal cancer, carcinoembryonic antigen is recommended (with some caveats) for prognosis determination, post-operative surveillance, and therapy monitoring in advanced disease. Fecal occult blood testing is considered useful for screening asymptomatic adults who are older than 50 years.
Breast cancer
For breast cancer, estrogen and progesterone receptors predict response to hormone therapy, human epidermal growth factor receptor-2 predicts response to trastuzumab, while urokinase plasminogen activator/ plasminogen activator inhibitor 1 is used for determining prognosis in lymph node-negative patients. CA15-3/BR27–29 or carcinoembryonic antigen can be used for therapy monitoring in advanced disease.
Ovarian cancer
CA125 is recommended (with transvaginal ultrasound) for early detection of ovarian cancer in women at high risk for this disease. CA125 is also recommended for differential diagnosis of suspicious pelvic masses in post-menopausal women, as well as for detection of recurrence, monitoring of therapy, and determination of prognosis in women with ovarian cancer.
Future research to target higher specificity and sensitivity
In the future, research is expected to focus on finding markers which are specific of one pathology, have higher sensitivity with a low cut-off and deliver results which correlate to tumour mass and growth potential. Ideal candidates would also have a short life duration to permit efficient follow-up; in other words, their presence should decrease during treatment and increase before a relapse.
The promise and challenges of screening
Given that tumour markers can aid in assessing the response to cancer treatment and making prognoses, many public health professionals have hoped they might also be used for screening tests which would detect cancer before the presence of symptoms.
Indeed, many tests have both screening and diagnostic uses, with only the context of use determining whether the test is one or the other. “A screening test is done on asymptomatic individuals who receive the test principally because they are of the age or sex at risk for the cancer. A diagnostic test is done on an individual because of clinical suspicion of disease.”
However, no tumour marker identified to date is sufficiently sensitive or specific to be used on its own for screening, demonstrating a survival benefit in randomized controlled trials in the general population.
One of the best known examples is the prostate-specific antigen (PSA) test. Although it is now accepted that most men with elevated PSA levels do not have prostate cancer, the implications of this remain mired in controversy and also illustrate the kind of limitations which other tumour biomarkers may face in the future for use in screening.
PSA screening has been the subject of two large randomized controlled trials in the US and Europe in the 2000s. However, as the Mayo Clinic notes, in spite of the size of the trials, there were “no clear conclusions.” This is because their diversity of methodology allows for significant flexibility in interpretation. As a result, “the decision of whether to screen or not screen – using PSA testing or other means or both – is a decision best made between physicians and their individual patients.”
The two trials were PLCO (Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial) conducted by the National Cancer Institute in the US, and the European Randomized Study of Screening for Prostate Cancer (ERSPC), billed as “the largest randomized trial of screening for prostate cancer” with 162,388 subjects.
In 2011, a US government task-force concluded that healthy men should not be screened for prostate cancer. The finding, which “drastically changed the standard of care for middle-age American men who had grown accustomed to annual screenings,” was largely based on 10 years data from the two studies, which found risk of over-diagnosis and over-treatment.
Problems began when follow-on data from ERSPC two years later showed screening was associated “with a 21% reduction in risk of prostate cancer mortality.” However, this was accompanied by a still-sizeable risk of over-diagnosis and over-treatment. As a result, the authors said that “population-based screening could not yet be recommended.
In April 2015, an article in ‘The Lancet’ re-confirmed “a substantial 21% reduction” from PSA screening. However, due to access restrictions to the ERSPC trial data, the authors called this figure into serious question.
The controversy is unlikely to go away for some time. An Op Ed in The New York Times called the PSA test “hardly more effective than a coin toss.” Although the date of publication was 2010, the author of the commentary was Dr. Richard Ablin, who discovered PSA in 1970.
Such challenges are also likely to accompany screening for other conditions. For instance, data from the PLCO trial show that screening for CA-125 (recommended by the National Academy of Clinical Biochemistry for women with ovarian cancer, along with transvaginal ultrasound), does not reduce ovarian cancer mortality. Instead, false-positive screening test results have been associated with complications.