Repeated Blood Test Patterns Are Emerging as a Cancer Detection Signal
Repeated blood tests are no longer being examined only as static snapshots. In this scoping review, investigators mapped how blood test trends have been used in cancer detection, and the literature now looks broad enough to show a real signal, yet fragmented enough to resist firm clinical translation. Ninety studies met inclusion criteria, spanning published and unpublished work identified through MEDLINE, Embase, ProQuest Dissertations & Theses Global, and Overton searches conducted from inception to April 2025.
The central idea is clinically intuitive. Malignancy often alters physiology gradually, before a diagnosis is ever coded, and serial laboratory values may capture those slow changes better than a single abnormal result. A falling haemoglobin trajectory, a shifting HbA1c pattern, or other repeated laboratory deviations may reflect occult disease biology, chronic inflammation, nutritional disruption, or paraneoplastic effects long before imaging or symptoms trigger a work-up. The review catalogued how researchers have tried to turn those patterns into earlier recognition of cancer rather than relying on threshold-based abnormality alone.
Colorectal and Pancreatic Cancer Drove Most of the Evidence Base
The literature was concentrated in two cancers. Colorectal cancer and pancreatic cancer were the most commonly studied malignancies, with haemoglobin emerging as the most frequently investigated test in colorectal cancer and HbA1c the most frequently investigated test in pancreatic cancer. Those selections make biological sense. Colorectal cancer can produce chronic blood loss and evolving anaemia, while pancreatic cancer can disrupt glycaemic control through pancreatic injury, insulin resistance, and new-onset dysregulation that may precede diagnosis.
Research interest, however, was not evenly distributed across the cancer spectrum. Many other tumour types were underrepresented, leaving the field weighted toward a small number of clinical scenarios that are more plausible or easier to study using routinely collected blood data. The resulting evidence base is informative, but it does not yet support broad claims that serial blood testing performs similarly across cancers with very different biology, symptom trajectories, and health-system pathways.
Association Studies Outnumbered Predictive Modelling Efforts
Most studies focused on whether blood test trends were associated with cancer at all. Sixty-four studies examined the association between laboratory trajectories and cancer, while 26 evaluated prediction models that incorporated blood test trends to estimate cancer risk. The balance matters because association is only the first step. A variable can be statistically linked to cancer without being useful for triage, risk stratification, or referral decisions in daily practice.
Prediction modelling brings the field closer to implementation, but it also raises a higher bar. A usable model must perform well in the messy conditions of primary care, where serial testing frequency varies, comorbidity is common, and not every patient follows a neat observation schedule. The review found that many studies were still working at the exploratory stage, where the main question is whether trends contain information, not yet whether they can safely alter referral thresholds or reduce diagnostic delay.
Trend Analysis Was Often Combined With Other Risk Signals
Several thematic areas appeared repeatedly across the included studies. One large cluster involved combining trends with other cancer risk factors, reported in 62 studies. Another examined practice implications in 58 studies. Forty studies compared trends with abnormality alone, 39 studied combinations of trends across different blood tests, and 12 addressed bias from informative observation.
Those themes reflect a field trying to answer a practical question: does longitudinal information improve on a single abnormal value? In many cancers, the answer may depend on context. A modest decline in haemoglobin may be more meaningful when paired with age, smoking history, weight loss, or inflammatory markers. Likewise, a changing test pattern may become more informative when multiple laboratory signals move together rather than in isolation. Serial data can strengthen suspicion, but they can also mislead if clinicians order tests more often in people already perceived to be at higher risk, which is exactly where informative observation bias becomes clinically important.
Serial testing can sharpen suspicion, but it can also magnify bias when sicker patients are monitored more often than healthier ones.
Informative Observation Remains a Key Methodological Problem
The review identified informative observation as an explicit theme in only 12 studies, which is a reminder that methodologic rigor has not kept pace with enthusiasm. In routine care, blood tests are not drawn at random intervals. Patients with symptoms, comorbid disease, or clinician concern are tested more often, and that pattern can distort any attempt to use trends as a detection tool. A downward haemoglobin slope may reflect an underlying cancer, but it may also reflect repeated measurement in a patient already under closer scrutiny for noncancer reasons.
This is the point at which prediction science needs cleaner epidemiology. Trend-based models must account for why tests were ordered, how often they were repeated, and whether the observed change is truly new or simply the result of more intense surveillance. Without that correction, apparent predictive value can be inflated, especially in retrospective datasets where timing is determined by clinical behaviour rather than protocol.
The Field Is Promising, but the Implementation Question Is Still Open
The review concludes that evidence for blood test trends in cancer detection is expanding, but remains fragmented and largely exploratory. That assessment fits the current state of the literature. Researchers have shown enough consistency to justify continued work, especially in colorectal and pancreatic cancer, yet the field still lacks broad validation across other malignancies, consistent modelling strategies, and practical implementation studies that show how trend information should change care.
What clinicians need next is not more proof that laboratory trajectories can correlate with cancer in selected datasets. They need evidence that these patterns improve early detection without flooding pathways with false positives, unnecessary repeat testing, or avoidable referrals. They also need clarity on whether particular trends matter more in specific cancers, whether combining multiple blood tests improves performance, and how serial laboratory data should be interpreted alongside symptoms and established risk factors.
For now, the most defensible reading is cautious. Repeated blood tests may contain useful early warning information, especially where physiology changes before diagnosis becomes obvious. The challenge is moving from signal to strategy, and from strategy to a detection tool that works in ordinary clinical practice rather than only in retrospective analyses and curated research cohorts.