Swedish Screening Model Improves Performance of Prostate Cancer Screening vs PSA Alone in Men Aged 50–69 Years


Key Points

  • The Stockholm 3 study model improved the AUC for diagnosis of high-risk cancer vs prostate-specific antigen alone from 0.56 to 0.74.
  • Use of the model could reduce the number of biopsies by 32% and allow 44% of benign biopsies to be avoided.

In a Swedish study reported in The Lancet Oncology (Stockholm 3 study), Grönberg et al found that a screening model including prostate-specific antigen (PSA) and other plasma protein markers, genetic polymorphisms, and clinical variables improved the accuracy of detecting prostate cancer with a Gleason score of at least 7 compared with PSA testing alone in men aged 50–69 years.

Study Details

The study included 11,130 men in a training cohort and 47,688 in a validation cohort. The predefined Stockholm 3 study model consisted of PSA; free PSA; intact PSA; hK2; MSMB; MIC1; a total of 232 single-nucleotide polymorphisms combined in a genetic score; and the clinical variables of age, family history, previous prostate biopsy, and prostate exam.

The primary aim of the study was to increase the specificity of screening compared with PSA without decreasing sensitivity to high-risk prostate cancer. The primary outcomes were the number of detected high-risk cancers (sensitivity) and the number of performed prostate biopsies (specificity measure). The Stockholm 3 training cohort was used to train the Stockholm 3 study model, which was then prospectively evaluated in the validation cohort.

Screening Model Performance

All variables in the model were significantly associated (P < .05) with high-risk prostate cancers in a multiple logistic regression model. The Stockholm 3 study model performed significantly better than PSA alone in detection of cancers with a Gleason score of at least 7 (P < .0001), with an area under the curve (AUC) of 0.74 (95% confidence interval [CI] = 0.72–0.75) for the Stockholm 3 study model vs 0.56 (95% CI = 0.55–0.60) with PSA alone. In a stepwise AUC analysis, the model components of risk factors (age, family history, and biopsy history), a combined biomarker score (genetic score and plasma protein biomarkers), and prostate exam (digital rectal exam and prostate volume) independently improved test characteristics of the model (each P < .0001).

At the same level of sensitivity as PSA testing with a cutoff of ≥ 3 ng/mL to diagnose high-risk disease, use of the Stockholm 3 study model was predicted to reduce the number of biopsies by 32% (95% CI = 24%–39%) and to avoid 44% (95% CI = 35%–54%) of benign biopsies. Of 603 high-risk cancers identified using the Stockholm 3 study model, 124 (21%) were in a PSA range of 1 to 3 ng/mL. Use of the model reduced the number of cancers identified with a Gleason score of 6 by 17% vs PSA alone.

The investigators concluded: “The STHLM3 model could reduce unnecessary biopsies without compromising the ability to diagnose prostate cancer with a Gleason score of at least 7 and could be a step towards personalised risk-based prostate cancer diagnostic programmes.”

The study was funded by the Stockholm County Council.

Henrik Grönberg, MD, of the Karolinska Institutet, is the corresponding author of The Lancet Oncology article.

The content in this post has not been reviewed by the American Society of Clinical Oncology, Inc. (ASCO®) and does not necessarily reflect the ideas and opinions of ASCO®.