Research Foundation

Rethinking Pulmonary Arterial Pressure as a Bounded and Dynamic Phenotype in Cattle

Markel, C. D. · Holt, T. N. · Lake, S. L. · Engle, T. E. · Field, S. P. · Cunningham-Hollinger, H. C. · Gifford, C. L.
Department of Animal Science, University of Wyoming, Laramie, WY  ·  Department of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins, CO
Corresponding author: cmarkel1@uwyo.edu
University of Wyoming Colorado State University Pulmonary Physiology Genetic Evaluation

Abstract

Pulmonary arterial pressure (PAP) has been used for decades as a practical field phenotype to evaluate susceptibility to high-altitude pulmonary hypertension in cattle, yet its interpretation has largely remained tied to isolated raw scores and fixed threshold risk categories. This paper proposes a theoretical framework for PAP that treats the phenotype as a latent, repeated, and practically bounded physiological trait rather than as a static one-time measurement. Under this framework, observed PAP is conceptualized as the sum of an underlying latent state and random error, and repeated observations are used to improve estimation through shrinkage and mixed-model principles. A boundary-aware logit transformation is proposed to reduce scale compression near practical bounds. The broader implication is that PAP phenotyping should move beyond isolated scores toward baseline-plus-trajectory informed phenotypes that better capture underlying cardiopulmonary liability through time.

📐

Latent Variable Framework

Observed PAP = true latent state + measurement error. Single scores are incomplete summaries of the underlying physiology.

🔄

Shrinkage Estimation

Extreme observed values are pulled toward the population mean in proportion to their reliability — reducing the impact of noise.

📊

Boundary-Aware Scaling

A logit transformation maps bounded PAP (30–150 mmHg) to an unbounded latent scale, recovering information compressed near the edges.

📈

Trajectory-Informed Risk

Direction of change over time carries independent information beyond the current PAP score. Rising trajectories increase future risk.

Introduction

Pulmonary arterial pressure (PAP) is widely used in beef cattle as an indicator of cardiopulmonary adaptation and susceptibility to high-altitude pulmonary hypertension, or brisket disease. In hypoxic environments, pulmonary arteries constrict to redistribute blood flow, but in susceptible cattle this response can become excessive, increasing pulmonary vascular resistance, elevating PAP, and contributing to right ventricular hypertrophy and eventual right-sided failure.

The value of PAP extends beyond clinical screening because the trait is moderately heritable (0.26–0.34), supporting its use in genetic evaluation and breed-level selection programs. At the same time, PAP is not purely static — repeated-measures work has shown that PAP can increase with age, adiposity, production stage, and environmental context. This means a single yearling record may not always represent an animal's longer-term cardiopulmonary status.

This creates a statistical problem: any PAP record is an observed value, but the liability of real interest is the animal's underlying cardiopulmonary state. Extreme values tend to be followed by less extreme values even in the absence of true biological change — a phenomenon known as regression to the mean. Together, these ideas suggest that PAP should be treated not as a one-time score, but as a dynamic phenotype whose interpretation depends on repeated measurement, scale, and the distinction between observed values and the underlying process.

Theoretical Framework: Latent Variables and Measurement Error

Let the latent (true) pulmonary arterial pressure for animal i be denoted as μi — the underlying physiological PAP that would be observed if measurement were perfect. The observed PAP at time t is:

yi,t = μi + εi,t εi,t ~ N(0, σ²_ε) [measurement error, unbiased] μi ~ N(μ, σ²_μ) [population distribution of true PAP]

The key insight is that σ²_μ represents real biological heterogeneity among animals, while σ²_ε represents noise around each animal's true level. To estimate μi, the tool uses shrinkage estimation — the posterior mean of μi is a weighted average of the observed value and the population mean:

μ̂i = λ · yi + (1 − λ) · μ where λ = σ²_μ / (σ²_μ + σ²_ε) [shrinkage factor]

When measurement error is large, λ is small and the estimate is pulled toward the population mean. When the signal is reliable, λ approaches 1 and the observed value is trusted more heavily.

Boundary-Aware Logit Transformation

Because PAP is observed within a finite practical range (30–150 mmHg), stochastic movement near the boundaries is asymmetric — animals near the lower bound have more room to increase than decrease, and vice versa. This creates scale compression that makes animals clustered at the low end appear more similar than they biologically are.

The tool addresses this by rescaling observed PAP to the unit interval and applying a logit transformation:

p = (y − L) / (U − L) [rescale to [0,1]; L=30, U=150] z = log(p / (1 − p)) [logit: maps [0,1] → (−∞, +∞)]

This transformation maps the bounded interval to the real line, producing a scale on which changes can be modeled without the compression that occurs near the practical bounds. In practical terms, this helps distinguish whether a given PAP value reflects a truly low or high underlying physiological state, rather than simply where that animal falls near the natural limits of the observed scale.

Animals that appear clustered together on the raw PAP scale can become more distinguishable on the latent scale.
Biologically meaningful information is not confined to extreme high PAP values — the lower and intermediate ranges contain signal that is compressed on the raw scale.
Identical raw PAP values at a single time point can reflect different biological histories and trajectories.

Key Discussion Points

The central premise is that PAP should not be treated as a simple static score. PAP is a repeated, physiologically constrained phenotype observed with error and expressed through time. Under that view, the conventional practice of interpreting one observed value as though it were the animal's complete and stable cardiopulmonary identity is mathematically incomplete.

This has direct consequences for comparing animals. On the raw scale, two animals with the same observed PAP are often treated as biologically comparable. But a value of 48 mmHg may not carry the same biological meaning in an animal with a low stable baseline and upward trajectory as it does in one with a historically higher baseline that is stable or declining.

The same logic extends to phenotype construction for genetic evaluation. If the goal of selection is to identify animals with superior underlying cardiopulmonary resilience, the phenotype submitted to the evaluation system should represent as much of that latent biology as possible. A trajectory-informed phenotype could potentially improve genetic prediction by placing more information into a single submitted record.

Implications

Field Screening

Risk characterization should move beyond threshold logic toward gradient-based assessment that accounts for baseline, trajectory, and position on an approximately unbounded scale.

Genetic Evaluation

A trajectory-informed, latent-scale phenotype may improve EPD accuracy by providing a richer summary of the underlying biology than a single raw score.

Repeated Testing

Multiple PAP records reduce the impact of measurement noise and provide trajectory information — improving estimation of the true latent state.

Interpretation

Low raw PAP values that appear clustered may still contain meaningful latent separation. Information is distributed across the full trait continuum.

Conclusion

PAP has long been treated as a useful but imperfect indicator trait. Many of its limitations arise not because the phenotype is fundamentally uninformative, but because it has generally been interpreted on a scale and in a form that does not fully match its biological structure. PAP is more appropriately viewed as a latent, time-dependent, and practically bounded physiological trait observed with measurement error.

A boundary-aware latent framework provides a mathematically coherent way to reinterpret this phenotype. By distinguishing observed PAP from the underlying physiological state, and by transforming the bounded observed scale to an approximately unbounded latent scale, the approach creates room to recover information that is compressed on the raw scale.

After roughly six decades of relying on PAP as a practical field phenotype, progress in its usefulness may depend less on replacing the trait than on handling it in a mathematically and biologically more appropriate way.

Literature Cited

Barnett, A. G., J. C. van der Pols, and A. J. Dobson. 2005. Regression to the mean: what it is and how to deal with it. Int. J. Epidemiol. 34:215–220.
Crawford, N. F., et al. 2016. Heritabilities and genetic correlations of pulmonary arterial pressure and performance traits in Angus cattle at high altitude. J. Anim. Sci. 94:4483–4490.
Fuller, W. A. 1987. Measurement Error Models. Wiley.
Henderson, C. R. 1975. Best linear unbiased estimation and prediction under a selection model. Biometrics 31:423–447.
Holt, T. N., and R. J. Callan. 2007. Pulmonary Arterial Pressure Testing for High Mountain Disease in Cattle. Vet. Clin. North Am. Food Anim. Pract. 23:575–596.
Neary, J. M., et al. 2015a. The altitude at which a calf is born and raised influences the rate at which mean pulmonary arterial pressure increases with age. J. Anim. Sci. 93:4714–4720.
Neary, J. M., et al. 2015b. Mean pulmonary arterial pressures in Angus steers increase from cow-calf to feedlot-finishing phases. J. Anim. Sci. 93:3854–3861.
Pauling, R. C., et al. 2018. Evaluation of moderate to high elevation effects on pulmonary arterial pressure measures in Angus cattle. J. Anim. Sci. 96:3599–3605.
Robinson, G. K. 1991. That BLUP is a Good Thing: The Estimation of Random Effects. Stat. Sci. 6.
Schaeffer, L. R. 2004. Application of random regression models in animal breeding. Livest. Prod. Sci. 86:35–45.
Shirley, K. L., D. W. Beckman, and D. J. Garrick. 2008. Inheritance of pulmonary arterial pressure in Angus cattle and its correlation with growth. J. Anim. Sci. 86:815–819.
Smithson, M., and J. Verkuilen. 2006. A better lemon squeezer? Maximum-likelihood regression with beta-distributed dependent variables. Psychol. Methods 11:54–71.
Speidel, S. E., et al. 2020. Evaluation of the sensitivity of pulmonary arterial pressure to elevation using a reaction norm model in Angus Cattle. J. Anim. Sci. 98:skaa129.
Go to Analyzer