How Does the PhenoAge Biological Age Calculator Work?
Summary ID01-BA-RS-00020
Research Summary
Original Publication
A New Aging Measure Captures Morbidity and Mortality Risk Across Diverse Subpopulations From NHANES IV: A Cohort Study
Journal
PLOS Medicine, .
- Category
- 01Biological Age
- Evidence
- Cohort study
- Summary Type
- Core
What Was Studied?
Researchers evaluated Phenotypic Age, or PhenoAge, a clinical biological-age measure calculated from nine routinely measured blood biomarkers together with chronological age.
The nine biomarkers used to calculate PhenoAge are albumin, creatinine, glucose, C-reactive protein (CRP), lymphocyte percentage, mean cell volume (MCV), red cell distribution width (RDW), alkaline phosphatase (ALP) and white blood cell count.
The final model therefore uses 10 variables in total: chronological age plus nine clinical biomarkers. These biomarkers had previously been selected using a Cox proportional hazards elastic-net model for mortality with 10-fold cross-validation. The Phenotypic Age algorithm is based on two Gompertz proportional hazards models.
Unlike biological-age methods developed primarily to reproduce chronological age, PhenoAge was designed to distinguish mortality risk between people of the same chronological age. The resulting value is expressed in years and represents the age within the general population corresponding to an individual's estimated mortality risk.
The researchers tested the method in an independent, nationally representative US population from NHANES IV. The analysis included 11,432 adults aged 20-84 years, together with 185 oldest-old adults whose ages were top-coded at 85 years. A total of 1,012 deaths were recorded over 12.6 years of follow-up.
The researchers examined whether Phenotypic Age was associated with mortality, disease burden and health outcomes across different ages, health states and population subgroups.
Key Findings
Phenotypic Age provided information about mortality risk beyond chronological age alone.
After adjustment for chronological age, each additional year of Phenotypic Age was associated with approximately 9% higher all-cause mortality risk.
The association remained across different age groups and after excluding deaths occurring during the first five years of follow-up, reducing the likelihood that the result simply reflected severe illness near the end of life.
Phenotypic Age also distinguished risk among people who appeared clinically healthy. Among 1,906 participants who reported no disease and had a normal BMI, each additional year of Phenotypic Age was associated with approximately 8% higher all-cause mortality risk.
The measure was also associated with disease burden: people with more chronic diseases tended to have a higher Phenotypic Age relative to their chronological age.
For 10-year mortality prediction, Phenotypic Age achieved an AUC of 0.88, compared with 0.86 for chronological age, and performed better than the individual clinical biomarkers and several conventional risk factors examined in the study.
When compared with the earlier Levine Biological Age, based on the Klemera-Doubal method, overall performance was broadly comparable. Phenotypic Age showed better predictive performance in the healthy subgroup, although the difference in 10-year mortality prediction in that smaller subgroup did not reach statistical significance.
Why It Matters for Longevity
PhenoAge illustrates an important principle in biological age calculation: routine laboratory measurements can provide more information when interpreted together than when considered only as separate test results.
The nine biomarkers do not independently represent nine measures of aging. They are combined with chronological age within a defined statistical model developed around mortality risk.
A Phenotypic Age of 55 in a chronologically 50-year-old person does not mean that every organ or tissue is biologically 55 years old. It indicates that, according to the model, the person's clinical biomarker profile corresponds to a mortality risk observed on average in someone approximately 55 years old in the reference population.
That distinction is fundamental when biological-age calculations are presented directly to patients.
For practical use, the calculation itself must also be implemented precisely. In 2019, PLOS Medicine published a formal correction to the Phenotypic Age equation presented in the original article because one computational step had been omitted. The correction provided the complete equation required to calculate Phenotypic Age.
Clinical Perspective
PhenoAge, also frequently referred to as the Levine PhenoAge method, has several characteristics that make it particularly relevant to the development of practical Biological Age Calculators.
It uses routine clinical laboratory measurements, produces a result expressed intuitively in years and was developed around a meaningful health outcome rather than simply reproducing chronological age.
The study also provides evidence that the measure contains information beyond known disease status. Its association with mortality persisted among participants considered healthy and across multiple population subgroups.
At the same time, PhenoAge is a risk-derived biological-age estimate, not a direct measurement of the aging process itself.
The study was observational, and Phenotypic Age was measured at one point in time. The authors therefore could not determine whether changes in Phenotypic Age over time reflected changes in the underlying rate of aging or predicted subsequent disease accumulation.
This distinction becomes especially relevant when biological-age calculators are used repeatedly. A change in PhenoAge should not automatically be interpreted as evidence that the underlying rate of biological aging has accelerated or slowed.
For clinical or digital implementation, PhenoAge must be calculated using the complete published algorithm, with the required biomarkers, laboratory units and mathematical transformations correctly specified. The 2019 corrected equation should be used for implementation.
A calculator may appear simple at the user interface while still depending on a precisely defined underlying model.
Key Takeaway
PhenoAge shows how nine routine blood biomarkers can transform a standard laboratory profile into a biological-age estimate with meaningful prognostic information. People of the same chronological age can have different PhenoAge results and substantially different health and mortality risks.
Reviewed and Summarized by
Source Publication
- Title:
- A New Aging Measure Captures Morbidity and Mortality Risk Across Diverse Subpopulations From NHANES IV: A Cohort Study
- Authors:
- Zuyun Liu, Pei-Lun Kuo, Steve Horvath, Eileen Crimmins, Luigi Ferrucci, Morgan Levine
- Journal:
- PLOS Medicine, December 2018; 15(12):e1002718.
