What are the specificity and sensitivity of the Galleri multi‑cancer early detection test (GRAIL)?

Galleri Multi-Cancer Early Detection Test Performance

The Galleri test demonstrates high specificity (98.4-99.5%) but limited sensitivity (20.8-66.3% overall, with particularly poor performance for early-stage cancers at <20% for Stage I disease), making it unsuitable as a standalone screening tool for early cancer detection in the general population. 1

Test Accuracy Data

Specificity

• Consistently high across studies: 98.4-99.5% 1
• Real-world data from 111,080 individuals confirmed specificity with a cancer signal detection rate of only 0.91% (1,011/111,080), consistent with low false-positive rates 2
• The 99.5% specificity reported in clinical trials translates to approximately 0.5% false-positive rate 3

Sensitivity

The sensitivity varies dramatically by cancer stage and is the test's primary limitation:

• Overall sensitivity: 20.8-66.3% across three studies 1
• Stage I cancers: <20% sensitivity 3
• Stage I-II cancers combined: <30% sensitivity 3
• Stage III-IV cancers: Higher sensitivity (specific percentages not consistently reported, but substantially better than early stages) 1

Cancer Signal Origin (CSO) Prediction

When the test detects a cancer signal, it performs well at predicting the anatomical location:

• 87% accuracy in predicting the correct cancer site of origin in real-world data from 258 confirmed cancer diagnoses spanning 32 cancer types 2
• This CSO prediction enabled efficient diagnostic workup with a median of 39.5 days from result receipt to cancer diagnosis 2

Critical Limitations for Screening

Biological Constraints

The fundamental problem is tumor burden and mutant allele fraction (MAF):

• Tumors must be approximately 10-15 mm in diameter to be reliably detected 3
• At this size, MAF is approximately 0.01% (one tumor DNA molecule per 10,000 normal DNA molecules) 3
• Using 10 mL of blood (4 mL plasma) may contain less than a complete cancer genome for very early cancers, rendering detection impossible 3

Clinical Implications

• The test misses the majority of early-stage cancers when intervention would be most beneficial 3
• While specificity is excellent, the poor sensitivity for Stage I disease means most curable cancers will not be detected 1, 3
• The test performs better for later-stage cancers, but these are often already symptomatic and would be diagnosed through standard clinical evaluation 1

Real-World Performance Context

From the largest real-world dataset available:

• Of 1,011 individuals with cancer signals detected, only 459 had reported clinical outcomes 2
• Of these 459, only 258 had confirmed invasive cancer diagnoses 2
• This suggests that even with high specificity, the positive predictive value in real-world screening populations requires further validation 2

Comparison to Established Screening

For context, established cancer screening tests demonstrate superior sensitivity for early disease:

• Mammography for breast cancer: 77-95% sensitivity 4
• FIT for colorectal cancer: 74% sensitivity for cancer, 23% for advanced adenomas 4
• Ultrasound plus AFP for HCC: 74% sensitivity for early-stage disease 4

Bottom Line for Clinical Practice

The Galleri test should not be used as a primary screening tool for early cancer detection given its <20% sensitivity for Stage I cancers 3. The high specificity is reassuring for minimizing false positives, but the test fundamentally fails at its intended purpose of detecting cancers when they are most treatable 1, 3. Ongoing randomized controlled trials are needed to determine if any mortality benefit exists despite the poor early-stage sensitivity 1.