Objective Biomarkers for Total Added Sugar Intake - Are We on a Wild Goose Chase?

Abstract

Misreporting of added sugar intake has been the major criticism of studies linking high added sugar consumption to adverse health outcomes. Despite the advancement in dietary assessment methodologies, the bias introduced by self-reporting can never be completely eliminated. The search for an objective biomarker for total added sugar intake has therefore been a topic of interest. In this article, the reasons this search may be a wild goose chase will be outlined and discussed. The limitations and inability of the 2 candidate biomarkers, namely urinary sucrose and fructose and δ¹³C isotope, which are based on the 2 only possible ways (i.e., difference in metabolism and plant sources) to identify added sugar based on current knowledge in human physiology and food and nutritional sciences, are discussed in detail. Validation studies have shown that these 2 candidate biomarkers are unlikely to be suitable for use as a predictive or calibration biomarker for total added sugar intake. Unless advancement in our understanding in human physiology and food and nutritional sciences leads to new potential ways to distinguish between naturally occurring and added sugars, it is extremely unlikely that any accurate objective added sugar biomarker could be found. It may be time to stop the futile effort in searching for such a biomarker, and resources may be better spent on further improving and innovating dietary assessment methods to minimize the bias introduced by self-reporting.

Keywords: added sugars; biomarker; metabolite; urinary sucrose; validity; δ¹³C isotope.

FIGURE 1

(A) Scatterplot of total sugar vs. added sugar intake; and (B) distribution of proportion of total sugars as added sugars of adult respondents of the 2011–2012 Australian Health Survey (32). The line of best fit and R² in panel A were calculated using simple linear regression between x and y values. Panel A showed that there was great variability in the proportion of total sugars as added sugar across the spectrum of total sugar intake; and panel B illustrated that the distribution of proportion of total sugars as added sugar roughly follows a normal distribution.

FIGURE 2

(A) Replot of data where both axes have the same scale; and (B) scatterplot of nonnatural logged values. Data were obtained from reverse-engineering of Figure 2 in Hedrick et al. (49) (n = 106, Southwest Virginian adults). The lines of best fit and R² values were calculated using simple linear regression between x and y values. It was evident from the graphs that the predicted added sugar intake varied within a much smaller range (∼50 to 100 g/d) compared with the reported added sugar intake (∼10 to 250 g/d).

FIGURE 3

Bland–Altman plot of data obtained from reverse-engineering of Figure 2 in Hedrick et al. (49) (n = 106, Southwest Virginian adults). The solid horizontal line represents the mean difference between the predicted and reported added sugar intake, and the 2 dotted horizontal lines represent the upper and lower limits of agreement (mean difference ± 1.96 × SD). The line of best fit and R² were generated using simple linear regression between x and y values. Results indicated there is a significant increase in underestimation as added sugar intake increases.