PSI-21 Optimizing sample size for accurate measurements and prediction equations for intestinal morphology in broiler chickens

Abstract

Abstract Histological measurements of intestinal morphology, including villi height (VH) and crypt depth (CD), are routinely used to evaluate intestinal health and the nutrient absorptive capacity in monogastric animals. These metrics suffer from significant natural variation, and there is no established standard for the number of unique measurements required to robustly estimate these variables for a given animal. Furthermore, the potential utility of additional variables, including villus width (VW), perimeter (VP), or villus area (VA), has not been explored. Therefore, the objective of the current experiment was to estimate the minimum number of measurements required to control the intrasample variance in each variable and determine if more complex or time-consuming metrics, such as VP and/or VA, can be accurately estimated from linear measurements. A set of 1283 intestinal morphology data, including VH, CD, VW, VP, and VA, measured from 68 intestinal tissues (46 ileum and 22 jejunum) of 4 experiments were used in the current study (Figure 1). Retrospective calculation of the CV for all combinations of N measurements per sample was used to determine the impact of measurement number on the quality of data. Data for the jejunum and ileum were pooled together, and the NLIN procedure of SAS was used to estimate the optimal measurement for intestinal morphology measurements using a quadratic broken line. The VA was estimated using geometric calculation with the assumption that VA equals 0.67 × base width × VH. A paired t-test was conducted to compare measured and calculated VA. The optimal number of measurements to minimize within-sample CV for VH, CD, VW, VP, and VA ranged from 10.6 to 10.8. The VH was highly correlated (P < 0.05) with VP (r = 0.95 and 0.87) and VA (r = 0.77 and 0.63) in the jejunum and ileum, respectively (Table 1 and 2). Additionally, there was a significant difference (P < 0.05) between measured (0.110 µm2) and calculated VA (0.086 µm2). In conclusion, this study shows that measuring 11 villi per sample is required for minimizing variations within the jejunal or ileal tissue in broiler chickens. Also, estimating VA based on idealized geometry was found to be inaccurate, necessitating either direct measurement or more complex prediction equations.