Diet quality and nutritional adequacy during a 2-year calorie restriction intervention: the CALERIE^™ 2 trial

Abstract

Background:

Comprehensive Assessment of Long-Term Effects of Reducing Intake of Energy (CALERIE^™) was the first randomized controlled trial of calorie restriction (CR) on biomarkers of aging and cardiometabolic health in humans without obesity.

Objective:

The aim of this secondary data analysis was to evaluate diet quality and nutritional adequacy during a 2 y CR intervention among healthy adults in the CALERIE^™ trial.

Methods:

CALERIE^™ 2 was a multi-site trial of healthy adults randomized to 2 y of 25% CR or an ad libitum (AL) control condition. CR participants received extensive dietary education and support. Food records at baseline and months 6, 12, 18, and 24 were analyzed using Nutrition Data System for Research. Diet quality was evaluated using the Probability of Adequate Nutrient Intake (PANDiet) diet quality index, Healthy Eating Index (HEI), and Dietary Inflammatory Index (DII). Nutritional adequacy was defined using sex- and age-specific Estimated Average Requirement or Adequate Intake criteria for each nutrient.

Results:

218 participants began the trial and are included in the analyses (143 CR, 75 AL; 69.7% females, age 38.1±7.2 y (mean±SD), baseline BMI 25.1±1.7 kg/m²). 188 participants completed the trial (82% CR, 95% AL). Average CR achieved during the 2 y intervention was 11.9±7.2%. Diet quality scores improved during CR according to all three metrics, both within group and when compared to AL (all P<0.01): PANDiet (CR baseline 76.27%, 95% CI [75.45, 77.09]; CR 2 y average 77.38% [76.58, 78.19]), HEI (baseline 59.40 [57.62, 61.18]; 2 y average 66.83 [65.11, 68.55]), and DII (baseline −0.28 [−0.58, 0.01]; 2 y average −1.05 [−1.35, −0.74]. Nutritional adequacy was not compromised during the CR intervention.

Conclusion:

Diet quality improved and nutritional adequacy was maintained during a 2 y moderate CR intervention designed to enhance healthspan and comprised of comprehensive nutrition counseling among healthy adults without obesity.

Introduction

Since the 1950s, the reduction of calorie intake below energy requirements, referred to as calorie restriction (CR), has been shown to play a role in delaying age-related physiological decline while increasing longevity among various species (1, 2). As evidence continues to emerge highlighting the benefits of CR on biomarkers of aging and age-related pathologies, the interest in and importance of studying the effects of long-term CR on human health increases (3–6).

Coincident with the priority of evaluating the physiological effects of sustained CR in humans is the importance of the quality of the CR diet to optimize the anticipated benefits. As emphasized in the Scientific Report of the 2020 Dietary Guidelines Advisory Committee (7), high diet quality can have protective effects on health and wellness, while low diet quality can influence health negatively. A systematic review and meta-analysis of cohort studies involving 1,670,179 participants in 68 reports revealed wide-ranging health benefits of higher diet quality, as assessed by the Healthy Eating Index (HEI), Alternate Healthy Eating Index, and Dietary Approaches to Stop Hypertension scores (8). Healthspan benefits included reduced risks for cardiovascular disease, cancer, type 2 diabetes, and neurodegenerative diseases, while lifespan benefits were evident by lower risk of all-cause mortality. An updated systematic review and meta-analysis by the same authors (9), which included 3,277,684 participants in 113 reports, confirmed the earlier associations of higher diet quality with lower disease incidence and lower all-cause mortality. An analysis of data from 47,994 women in the Nurses' Health Study and 25,745 men in the Health Professionals Follow-up Study provides additional evidence for the protective effects of high diet quality on mortality and the risks associated with low diet quality (10). Etiological explanations for these relationships were provided by a study in which higher diet quality scores were associated with higher blood micronutrient levels and more favorable concentrations of biomarkers associated with inflammation and metabolic health (11). These healthspan and lifespan effects highlight the importance of ensuring that CR diets do not jeopardize diet quality. Rather, individuals who make the effort to follow a CR diet for its potential to enhance healthspan or to slow aging likely aim to derive maximal benefits from such efforts.

While diet quality may not be affected adversely by a CR diet, the reduction in energy intake that characterizes CR may compromise nutrient adequacy and pose risks for nutrient insufficiencies or deficiencies, particularly for those with lower energy requirements due to body size or age (12). The importance of nutrient adequacy was highlighted by an analysis of dietary data from 20,602 adults aged 30 years and older in the 1999–2010 National Health and Nutrition Examination Surveys (NHANES), which revealed that higher intake of several nutrients (i.e., dietary fiber, calcium, iron, magnesium, potassium, and vitamins A and E) was associated with lower mortality during a median 9.3 y of follow-up (13). Interestingly but not surprisingly, data from NHANES 2013–2016 indicate that many American adults are not meeting the dietary recommendations for fiber, several vitamins, and several minerals, while exceeding the recommended intake levels for sodium and saturated fat (14). Decreases in energy intake with CR may exacerbate the challenge of meeting micronutrient requirements and widen the gap between recommended and consumed amounts.

Clinical trials involving CR generally include dietary education and guidance by registered dietitians to enhance diet quality and ensure sufficient dietary micronutrient intake; vitamin and mineral supplements also may be provided to reduce the potential risk of nutrient insufficiencies. The Comprehensive Assessment of Long-Term Effects of Reducing Intake of Energy (CALERIE^™) trial was the first randomized controlled trial of calorie restriction on biomarkers of aging and cardiometabolic health in humans without obesity (http://calerie.duke.edu). Previously, we reported diet quality results from the CALERIE^™ phase 1 studies (15), in which CR interventions of 6 months to 1 y in duration and CR levels ranging from 10% to 30% did not compromise diet quality or nutritional adequacy. In the phase 1 studies, the foods were either self-selected or involved a combination of meal provision and self-selection. The CALERIE^™ 2 multi-center trial involved a 2 y CR intervention, during which dietary intake was predominantly self-selected. The aim of the current data analysis was to evaluate diet quality and nutritional adequacy during the CR intervention in the CALERIE^™ 2 trial.

Methods

Study Design and Population

CALERIE^™ 2 was a randomized controlled trial that assessed the effects of 2 y of 25% CR, in comparison with an ad libitum control condition, on biomarkers associated with aging and healthspan. Details of the trial design were published previously (16). The ClinicalTrials.gov Identifier is NCT00427193; the clinical trial was conducted from 2007 to 2012. Three clinical study sites included Pennington Biomedical Research Center (PBRC) in Baton Rouge, LA; the Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University (Tufts) in Boston, MA; and Washington University School of Medicine (WashU) in St. Louis, MO. The Duke Clinical Research Institute in Durham, NC served as the coordinating center. The study was approved by the Institutional Review Board at each clinical site and all participants provided written informed consent.

Healthy premenopausal females aged 21–47 y and males aged 21–50 y with a body mass index (BMI) ≥22.0 and <28.0 kg/m² were eligible for enrollment. Exclusion criteria included pregnancy, cardiometabolic or other disease, abnormal laboratory value on routine blood chemistries, psychiatric or behavioral problems, disordered eating, regular use of medications other than oral contraceptives, and engaging in exercise for 150 minutes/week or more.

Randomization

After baseline assessments were completed, participants were randomized to either the calorie restriction (CR) group, which was prescribed a 25% CR diet for 2 y, or the ad libitum (AL) control group, which was instructed to maintain their usual dietary intake. Randomization was designed to favor the CR group with a 2:1 allocation (2 CR to 1 AL), generated by the coordinating center using a permuted block technique, with stratification by study site (PBRC, Tufts, WashU), biological sex (female, male), and BMI (<25.0 kg/m², ≥25.0 kg/m²).

CR Intervention

Participants in the CR group were given an individualized calorie prescription (kcal/day) that equated to 25% CR (i.e., 25% reduction in daily energy intake). The prescription was calculated as 25% less than total energy expenditure (TEE) determined using the doubly labeled water (DLW) method during two consecutive 14-day periods of weight maintenance. Upon initiation of the 2 y CR intervention, CR participants were provided all meals and beverages for 27 days and were instructed to consume only the provided items, which met each participant’s 25% CR prescription. Three dietary patterns were offered, each for nine days: 1) Mediterranean, 2) low fat, and 3) low glycemic load. The goals were to provide the correct energy content with high diet quality and nutritional adequacy, to familiarize participants with the appropriate portion sizes, and to introduce three different dietary patterns. After the first 27 days, participants followed a self-selected diet and were able to choose the dietary pattern and diet composition that they preferred, with macronutrient guidelines in accordance with the Dietary Reference Intakes (17). This flexibility was intended to promote adherence. CR participants were provided with food scales.

Experienced registered dietitians and clinical psychologists provided individual and group education and counseling sessions for participants throughout the study. This approach has been effective in previous trials, including CALERIE^™ Phase 1 (18–20), the Look AHEAD trial (21), and the Diabetes Prevention Program (22). Counseling sessions occurred weekly for the first month, bi-monthly during months 2 through 12, and monthly during year 2. Additional individual sessions were provided as needed to promote adherence. Group counseling sessions occurred bi-weekly during the first 26 weeks of the intervention and then monthly from week 27 until the end of the 2 y intervention. Counseling sessions covered topics such as maintaining motivation, managing food cravings, goal setting, and social support. Individualized strategies and toolbox options were used to optimize adherence to the CR regimen (23). Participants were provided food scales, measuring cups, and measuring spoons to promote accurate quantification of portion sizes and calorie content. CR participants also received a daily multivitamin/mineral supplement (Nature Made Multi Complete) and a calcium supplement (Optimum Calcium Citrate) to prevent micronutrient insufficiencies and to minimize bone loss. Additional details of the intervention have been published previously (16, 24).

AL Control Condition

Participants in the AL group did not receive an intervention, dietary counseling sessions, or food scales; they did receive the multivitamin/mineral and calcium supplements that were provided to the CR group.

Assessment of Nutrient Intake

All participants completed 6-day food records on 2 occasions during the baseline period and during 1-week periods at months 12 and 24. CR participants also completed 6-day food records at months 6 and 18. Participants were instructed to record, in detail, all food and beverages consumed during the 6 days, including brand name, portion size (i.e., weight, volume, or dimensions of each item), preparation method, and all ingredients for recipes. All participants received measuring cups and measuring spoons to facilitate accurate quantification and were encouraged to bring in package labels. A registered dietitian reviewed the food record with each participant at the end of each 6-day recording period to ensure completeness and to add any missing details. All food records were analyzed in a blinded fashion at the University of Cincinnati, which served as the CALERIE^™ 2 Nutrition Reading Center, using Nutrition Data System for Research (NDSR) software (2010 version, developed by the Nutrition Coordinating Center at the University of Minnesota, Minneapolis, MN). The food record data did not include the multivitamin-mineral supplement or calcium supplement that were provided during the study.

Diet Quality

The primary outcome was diet quality assessed using three metrics: the Probability of Adequate Nutrient Intake (PANDiet) Diet Quality Index (25), the Healthy Eating Index (HEI-2015) (26), and the Dietary Inflammatory Index (DII) (27).

PANDiet provides a comprehensive assessment of both the overall adequacy and moderation of individual macronutrients, vitamins, and minerals. Importantly, the PANDiet index accounts for the number of days on which intake was recorded, as well as inter-day and within-group variability in average nutrient intake. The Adequacy and Moderation sub-scores of the PANDiet Index were calculated using the Institute of Medicine’s probabilistic approach (28). First, the probability of adequate intake of 26 nutrients (Supplemental Table 1) was calculated for each participant. Then, the Adequacy sub-score was calculated by averaging the probability of adequate intake for the CR and AL groups at each time point. Similarly, the probability of moderate intake for 17 nutrients (Supplemental Table 2) was calculated and a Moderation sub-score was determined for the CR and AL groups at each time point. Lastly, the composite PANDiet Index was calculated as the sum of the adequacy and moderation sub-scores by group and time point. Values range from 0 to 100%, with higher values reflecting higher diet quality.

HEI scores reflect how closely an individual’s dietary patterns align with the recommendations in the Dietary Guidelines for Americans. HEI-2015 scores were computed as described previously (29) based on the 2015–2020 Dietary Guidelines for Americans (30), coinciding with the timing of the CALERIE^™ 2 trial. The HEI score is calculated based on 13 components that include fruits, vegetables, greens and beans, whole grains, dairy, total protein, seafood, plant protein, fatty acids, and moderate intake of refined grains, sodium, added sugars, and saturated fats. HEI scores range from zero to 100, with a higher score indicating higher concordance with the Dietary Guidelines.

DII is a metric that reflects the inflammatory potential of the diet based on 28 of 45 food parameters that are thought to affect six inflammatory biomarkers (interleukin [IL]-1β, IL-4, IL-6, IL-10, tumor necrosis factor alpha, and C-reactive protein). The DII was applied as described previously (29, 31, 32). DII scores range from −9 to +8, with higher positive scores indicating a more pro-inflammatory diet, 0 reflecting a neutral diet, and negative scores indicating a non-inflammatory or even anti-inflammatory diet.

Nutritional adequacy based on EAR or AI

Secondary outcomes included nutritional adequacy for several macro- and micronutrients. The adequacy of each nutrient in the diet at baseline and during the 2 y CR and AL periods was determined for each participant at each time point relative to the sex- and age-specific EAR and AI values shown in Supplemental Table 1. Supplement intake was not included in this analysis, as we limited our analyses strictly to nutrients obtained from foods and beverages reported in the food records.

Dietary Patterns

The three dietary patterns that were provided at the beginning of the CR intervention (i.e., Mediterranean, low fat, and low glycemic load) were explored as tertiary outcomes to evaluate their potential influence on diet quality during the 2 y CR intervention. The alternate Mediterranean diet score (aMED) assesses adherence to a Mediterranean diet pattern (33). Low fat was defined as dietary fat intake <25% of total energy intake. Low glycemic load (GL) was defined by a GL value ≤10.

Percent CR

%CR was determined during each 6-month interval of the 2 y intervention in the CR group and during each 12-month interval in the AL control group. These 6- and 12-month %CR values were used in the analyses of diet quality for the corresponding time points. Average %CR throughout the 2 y trial was determined based on changes in fat mass and fat-free mass from baseline to the 2 y time point (34).

Statistical Analyses

Baseline descriptive statistics were computed for the CR and AL groups. The overall 2 y average value in the CR group was computed using the 4 CR intervention time points (months 6, 12, 18, and 24); the overall 2 y average value in the AL group was computed using the 2 AL control period time points (months 12 and 24). Multivariable linear mixed effects models were used to assess the associations between the two randomized groups and each diet quality index during the 2 y intervention; the models were adjusted for age, sex, randomized group, study visit, and study site as a random effect. All models included all available repeated measurements (baseline and months 6, 12, 18, and 24 for CR; baseline, month 12, and month 24 for AL) and a group-by-time interaction term. An additional analysis of the 2 y average (CR or AL condition) for each diet quality index relative to baseline was conducted using the modeling approach described above. Associations between average %CR achieved over the 2 y period and each dietary index were assessed using generalized linear models adjusted for age and sex, with separate models applied for the CR and AL groups. RM-ANOVA was used to compare the frequency of adequate intake for each nutrient (i.e., the proportion of participants meeting the age- and sex-specific EAR or AI) across time points in the CR and AL groups. An exploratory analysis was conducted to identify potential correlations between each dietary pattern and each diet quality metric. Statistical models did not adjust for potential under-reporting of energy or nutrient intake.

Based on the intention-to-treat principle, all participants who began their assigned condition were included in the analyses, regardless of whether they completed the CR intervention or AL condition. Missingness was at random and was unrelated to group randomization or the dietary outcomes evaluated; missingness was addressed through a linear mixed modeling approach. Analyses were performed using SAS 9.4 © (SAS Institute, Inc., Cary, NC). To control the false discovery rate, P values for multiple pairwise comparisons were adjusted within-model using the Benjamini-Hochberg procedure. An adjusted P value < 0.05 was considered statistically significant. Values are presented as percentages, mean ± standard deviation (SD), or mean and 95% Confidence Interval (CI), as indicated.

Results

Participants

As shown in Figure 1, 220 participants (145 CR, 75 AL) completed baseline assessments and were randomized to a study group, 218 participants (143 CR, 75 AL) began their assigned group, and 188 participants (117 CR, 71 AL) completed the 2 y CR or AL condition. Completion rates for the 218 participants who began the trial were 86% overall, 82% in the CR group, and 95% in the AL group. The analysis sample is comprised of the 218 participants (143 CR, 75 AL) who began their assigned group. Sample sizes for diet record data are shown in Figure 1. The month 24 sample size differs from the number of study completers because one CR participant who did not complete the intervention did complete the month 24 assessments, while three AL participants who completed the study did not have complete dietary data at month 24. Additionally, the AL group sample size for the 2 y averages (n=72) is larger than the sample size for month 12 (n=70) and month 24 (n=68) because some participants completed diet records at only one of these two time points and are included in the averages. The analysis sample includes 152 females (69.7%) and 66 males (30.3%) aged 38.1 years (SD 7.2, range 21.3–50.9) with an average BMI of 25.1 kg/m² (SD 1.7) at baseline. Participant characteristics by study group are shown in Table 1. There were no differences at baseline between the CR and AL groups for sex, age, or BMI.

Figure 1.

Consort diagram depicting participant enrollment, randomization, study completion, and available diet record data at each time point in the Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy (CALERIE^™) 2 trial.

Table 1.

Participant characteristics, energy expenditure, and energy intake in the calorie restriction (CR) intervention group and the ad libitum (AL) control group

	CR Group (n = 143)	AL Group (n = 75)
Characteristics
Sex, % Females	69.2%	70.7%
Age, baseline, y	38.2 ± 7.3	38.1 ± 6.9
BMI1, baseline, kg/m2	25.2 ± 1.8	25.1 ± 1.6
Total Energy Expenditure, kcal/day, determined by DLW
Baseline	2467 ± 406	2390 ± 385
Month 6	2220 ± 369
Month 12	2234 ± 394	2359 ± 409
Month 18	2204 ± 393
Month 24	2250 ± 430	2365 ± 405
% Calorie Restriction Achieved, determined by DLW2
Months 1 – 6	19.5 ± 8.7
Months 7 – 12	10.8 ± 8.1	1.3 ± 9.1%3
Months 13 – 18	8.8 ± 8.7
Months 19 – 24	8.2 ± 9.4	0.4 ± 9.1%3
Self-Reported Energy Intake, kcal/day, determined by 6-day food records
Baseline	2126 ± 559	2044 ± 481
Month 6	1795 ± 380
Month 12	1796 ± 408	1946 ± 497
Month 18	1797 ± 392
Month 24	1854 ± 469	1922 ± 465

Values in the table represent mean ± SD or percentage of sample (as indicated for sex).

¹BMI, body mass index;

²DLW, doubly labeled water.

³%CR values for the AL group reflect averages for months 1–12 and months 13–24. Sample sizes for the CR group at baseline and months 6, 12, 18, and 24 are 143, 137, 127, 120, and 118, respectively; sample sizes for AL at baseline and months 12 and 24 are 75, 70, and 68, respectively.

Energy Intake, and % CR

Energy intake at baseline, determined from total energy expenditure measured over 4 weeks using the DLW method, averaged 2467 ± 406 kcal/day in the CR group. A 25% reduction in daily energy intake (i.e., 25% CR) yielded a prescribed energy intake of 1847 ± 296 kcal/day throughout the 2 y CR intervention. %CR achieved by CR participants throughout the 2 y intervention averaged 11.9 ± 7.2%, with a range of −3.8% CR (reflecting an average increase in daily calorie intake of 3.8% relative to baseline intake) to +31.2% CR (reflecting an average decrease in daily calorie intake of 31.2% relative to baseline intake). As shown in Table 1, %CR was the highest during the first 6 months and lowest during the last 6 months of the 2 y intervention. Self-reported energy intake based on food records at each assessment time point is shown by group in Table 1. Supplemental Figure 1 shows baseline TEE and average energy intake for each CR participant during years 1 and 2 of the CR intervention, based on DLW results.

Diet Quality

Table 2 shows the adjusted means and 95% CIs for the three diet quality indices at baseline and each time point during the 2 y CR intervention and AL control period. Baseline diet quality scores did not differ between groups. During the 2 y CR intervention, diet quality improved for all three diet quality indices (i.e., PANDiet and HEI scores increased and DII scores decreased) in the CR group at each intervention time point relative to baseline and compared to the AL group. In contrast to the CR group results, the AL group did not demonstrate improvements in diet quality at any time point during the 2 y study period.

Table 2.

PANDiet, Healthy Eating Index (HEI), and Dietary Inflammatory Index (DII) scores at baseline and during 2 years of a calorie restriction (CR) intervention or an ad libitum (AL) control condition

Diet Quality Index	Calorie Restriction Group		Ad Libitum Group		P-value between groups	P-value for interaction
	mean [95% CI]	P-value	mean [95% CI]	P-value
PANDiet, %
Baseline	76.27 [75.45, 77.09]	--	75.10 [74.01, 76.19]	--	0.093
Month 6	77.40 [76.47, 78.33]	0.010	--	--	--
Month 12	77.53 [76.67, 78.39]	0.004	74.26 [73.12, 75.41]	0.131	0.004
Month 18	77.55 [76.60, 78.50]	0.004	--	--	--
Month 24	77.25 [76.34, 78.15]	0.013	74.41 [73.22, 75.61]	0.171	0.013	0.005
2 y average	77.38 [76.58, 78.19]	0.001	74.23 [73.16, 75.30]	0.070	0.001	0.0006
HEI
Baseline	59.40 [57.62, 61.18]	--	58.81 [56.41, 61.21]	--	0.758
Month 6	65.87 [63.82, 67.93]	<0.0001	--	--	--
Month 12	67.63 [65.68, 69.59]	<0.0001	58.95 [56.32, 61.58]	0.923	<0.0001
Month 18	67.29 [65.26, 69.33]	<0.0001	--	--	--
Month 24	67.14 [65.00, 69.28]	<0.0001	59.70 [56.85, 62.55]	0.648	0.0002	<0.0001
2 y average	66.83 [65.11, 68.55]	<0.0001	59.11 [56.79, 61.44]	0.729	<0.0001	<0.0001
DII
Baseline	−0.28 [−0.58, 0.01]	--	0.04 [−0.36, 0.44]	--	0.208
Month 6	−0.93 [−1.27, −0.60]	<0.0001	--	--	--
Month 12	−1.26 [−1.58, −0.93]	<0.0001	−0.06 [−0.50, 0.37]	0.588	0.0004
Month 18	−1.18 [−1.54, −0.82]	<0.0001	--	--	--
Month 24	−0.97 [−1.33, −0.62]	<0.0001	−0.27 [−0.74, 0.21]	0.130	0.130	0.001
2 y average	−1.05 [−1.35, −0.74]	<0.0001	−0.11 [−0.52, 0.30]	0.330	0.003	0.002

Mean and 95% confidence interval (CI) values were derived from a multivariable linear mixed model with adjustment for baseline age, sex, randomized group, study time point, group * time interaction term, and study site as a random effect. Within-group P-values reflect comparisons with baseline scores during the CR intervention or AL control periods. Comparisons for the 2 y average values were derived from separate linear mixed models for each index that included the average of months 6, 12, 18, and 24 for CR and months 12 and 24 for AL. P-values for all comparisons were adjusted using the Benjamini-Hochberg procedure and represent the differences relative to baseline values using a within-model approach. Sample sizes for the CR group at baseline and months 6, 12, 18, and 24 are 143, 137, 127, 120, and 118, respectively; sample sizes for AL at baseline and months 12 and 24 are 75, 70, and 68, respectively. Sample sizes for the 2 y averages are 137 for CR and 72 for AL.

Influence of %CR and Dietary Patterns on Diet Quality

Figure 2 shows the individual results for %CR achieved throughout the 2 y CR intervention or AL control condition versus the average 2 y PANDiet, HEI, and DII scores for females and males separately. In the CR group, %CR was not significantly associated with any diet quality metric (panels A, C, E). In the AL group (panels B, D, F), HEI was weakly associated with %CR. In addition, higher adherence to a Mediterranean diet pattern, lower fat intake, and higher GL were significantly correlated with higher dietary quality as measured by PANDiet (Pearson’s r = 0.699, −0.452, and 0.488, respectively), HEI (0.765, −0.437, and 0.085, respectively), and DII (−0.689, 0.341, and −0.498, respectively) during the 2 y CR period (all P < 0.0001). Similar relationships were observed in the AL group between these three dietary patterns and the three diet quality indices.

Figure 2.

Scatter plots showing % calorie restriction (CR) achieved versus diet quality based on Probability of Adequate Nutrient Intake (PANDiet) scores (panels A and B), Healthy Eating Index (HEI) scores (Panels C and D), and Dietary Inflammatory Index (DII) scores (Panels E and F) in the CR group (n=137) and the ad libitum (AL) control group (n=72). Data represent individual values for females (gold triangles) and males (blue circles) during the 2 y CR intervention and AL control periods. R² values and P-values represent the correlations of %CR and diet quality in each group, with females and males combined.

Nutritional Adequacy Based on EAR or AI

Figure 3 displays the proportion of participants who achieved 100% of EAR or AI for several vitamins, minerals, and other nutrients at baseline and at 2 y of the CR intervention (panel A) or AL control condition (panel B). At baseline, fewer than 70% of participants in both the CR and AL groups achieved adequacy for vitamins B5, C, D, E, and K; the minerals calcium, magnesium, and potassium; and fiber. During the 2 y CR intervention, the proportion of participants who achieved nutritional adequacy was equivalent to baseline or increased relative to baseline for most macronutrients and micronutrients. During CR, >90% of participants achieved 100% of EAR or AI for 6 of the 12 vitamins assessed (A, B1, B2, B3, B6, B12), 5 of 8 minerals (copper, iron, phosphorus, selenium, zinc), and 2 of 4 macronutrients (protein, carbohydrate). In contrast, ≤90% CR participants achieved adequacy for 6 vitamins (B5, B9, C, D, E, K), 3 minerals (calcium, magnesium, potassium), and 2 macronutrients (fiber and omega-3 fatty acids), although fiber increased relative to baseline. Fiber intake, expressed relative to daily energy intake, averaged 9.7 g/1000 kcal at baseline and 15.7 g/1000 kcal during the 2 y CR intervention. In the AL group, the intake of most nutrients did not change appreciably during the 2 y AL control period.

Figure 3.

Proportion of participants who achieved 100% of the Estimated Average Requirement (EAR) or Adequate Intake (AI) for vitamins, minerals, and other nutrients at baseline (BL, light gray bars) and during the 2 y calorie restriction intervention (CR, black bars) or ad libitum control condition (AL, black bars). Panel A, CR Group. Panel B, AL group. Numeric values above the bars represent the % of participants. Omega-3 FA = omega-3 fatty acids. Sample sizes for the CR group are 143 at baseline and 137 for the 2 y averages; sample sizes for AL are 75 at baseline and 72 for the 2 y averages.

For CR participants who did not achieve 100% of the EAR or AI during the CR intervention, we determined the percentage that they did achieve. As shown in Figure 4, CR participants who did not achieve 100% met >75% of the EAR/AI for 17 of 22 nutrients (i.e., vitamins A, B1, B2, B3, B5, B6, B9, B12; minerals calcium, copper, iron, magnesium, phosphorus, potassium, zinc; protein, omega-3 fatty acids), between 60–75% of the recommendation for 4 nutrients (vitamins C, E, K, and fiber), and <35% of the recommendation for vitamin D.

Figure 4.

Proportion of CR participants who did not achieve 100% of the Estimated Average Requirement (EAR) or Adequate Intake (AI) for vitamins, minerals, and other nutrients (Panel A) and the % EAR or % AI that those CR participants achieved for each nutrient (Panel B). Bars in both graphs represent values at baseline (BL, light gray bars) and average values during the 2 y calorie restriction intervention (CR, black bars). Numeric values above the bars represent the % of participants in Panel A and the % EAR or AI achieved in Panel B. Omega-3 FA = omega-3 fatty acids. Sample sizes are 143 at baseline and 137 for the 2 y averages.

Discussion

The Comprehensive Assessment of Long-Term Effects of Reducing Intake of Energy (CALERIE^™) 2 trial investigated the effects of CR on biomarkers of aging and cardiometabolic health in healthy adults without obesity. In the current analysis, we examined diet quality and nutritional adequacy during 2 y of moderate CR in the CALERIE^™ 2 trial and observed overall favorable results. Specifically, diet quality was higher in the CR group during the 2 y moderate CR intervention compared to baseline and compared to the AL control group, as evidenced by improvements in PANDiet, HEI, and DII scores. Additionally, nutritional adequacy was not compromised during the 2 y CR intervention.

Our findings align with and extend the results of CALERIE^™ Phase 1 pilot studies, in which CR interventions of 6 months and 1 y did not compromise diet quality or nutritional adequacy (15). The present study, encompassing a 2 y duration and additional diet quality metrics, provides valuable insights into the benefits and safety of sustained CR on diet quality and nutrient adequacy. It’s important to note that CR participants achieved an average of 11.9% CR during the 2 y intervention, which is lower than the prescribed 25% CR and can be deemed moderate. One explanation for the modest %CR achieved is that the individualized predictive body weight curves that were used to guide each participant’s weekly dietary goals and adherence throughout the intervention depicted a “zone of adherence” that was later determined to be inaccurate.(35) Specifically, the zone of adherence at the 2 y time point corresponded to 10.4% CR to 19.4% CR. On average, participants were within this zone based on 11.9% CR achieved. Therefore, we cannot determine whether a higher level of energy restriction would have yielded less favorable nutritional adequacy results; it is plausible that achievement of 25% CR consistently for 2 y or longer may compromise intake of specific nutrients. Interestingly, however, diet quality was not associated with the level of CR achieved in this trial, suggesting that greater adherence to the CR prescription may not compromise diet quality.

Long-term CR interventions are of great interest for their potential to increase healthspan and slow biologic aging. Therefore, it is critical to ensure that a high-quality diet with adequate nutrients can be achieved. It is likely that the comprehensive nutrition education provided by registered dietitians and continuous intervention support provided to CR group participants contributed to the improvements in indices of diet quality in this group. The observed improvements in diet quality combined without decrements in nutritional adequacy during the intervention underscore the value of a comprehensive program with frequent contact in promoting healthy dietary choices during a CR intervention.

We included three distinct metrics of diet quality to encompass various aspects that are important in an overall healthful diet. The PANDiet score is comprised of Adequacy and Moderation sub-scores, which are highly relevant during an energy-restricted diet that imposes a reduction in overall nutrient intake. Adequacy, in particular, may be compromised by virtue of a chronic energy deficit during a long-term CR intervention. In comparison to our CR and AL group PANDiet total scores of 76.27 (range: 63.73–87.24) and 75.10 (range: 61.02–85.37), respectively, at baseline, reported mean values in the literature were 58.73 (SEM: 0.36, range: 34.74–89.97) among 2,391 adults in the U.S. (25), 63.25 (SEM: 0.29, range: 42.69–89.61) among 1,330 adults in France (25), 63.69 (SEM: 0.23, range: 38.27–89.74) among 1,051 adults in the Irish National Adult Nutrition Survey (36), and 60.48 (SE: 0.33) and 58.58 (SE: 0.19) among 2,797 adults classified as yogurt consumers or non-consumers, respectively, in the Italian National Food Consumption Survey INRAN SCAI 2005–06 (37). These comparisons suggest that our highly-screened CALERIE^™ sample of healthy young to middle-aged adults had higher diet quality compared to larger cohorts of adults with a wider age range and varying health statuses. The modest improvement in the PANDiet total score observed in the CR group in our CALERIE^™ 2 trial may be explained by the relatively favorable PANDiet scores at baseline. The improvement in the PANDiet total score was driven primarily by an increase in the PANDiet Moderation sub-score, indicating that CR moderated the intake of carbohydrates, total fat, saturated fatty acids, cholesterol and sodium.

The second diet quality metric, HEI, is a widely used dietary index that assesses and quantifies the extent to which dietary patterns align with the Dietary Guidelines for Americans. HEI scores observed at baseline in the CR and AL groups (59.4 and 58.8, respectively) and during the 2 y control period in the AL group (59.1) are consistent with average U.S. HEI scores of 56–59 in similarly aged adults (38). During the 2 y CR intervention, the CR group achieved a higher average HEI score of 66.8, demonstrating that diet quality can be improved while following a moderate CR diet. The observed improvements in HEI scores during the CR intervention generally indicate higher intake of fruits, vegetables, leafy greens, beans, whole grains, dairy, and plant protein, and reduced intake of sodium, added sugars, saturated fat, and refined grains.

HEI is a clinically meaningful metric, with higher scores representing better diet quality and more favorable health outcomes. A cohort study of adults with hypertension reported an inverse relationship between HEI scores and all-cause and cardiovascular mortality (39). Furthermore, an analysis of individuals with cancer in NHANES III revealed that participants in the highest HEI score quartile had a significantly lower risk of all-cause and cancer-specific mortality compared to those in the lowest HEI score quartile (40). Similarly, another study reported 21% lower cancer-specific mortality in participants with high HEI scores, although differences were not observed for all-cause or cardiovascular mortality (41). One could posit that the observed increases in diet quality in the CR group in our CALERIE^™ 2 trial may have contributed to the reported improvements in markers of cardiometabolic health and healthspan (42–46). A previous report indicated that HEI in the CALERIE^™ 2 trial was associated inversely with the metabolic syndrome score (29).

Our third diet quality metric, the DII, is designed to assess the inflammatory nature of an individual’s diet, with scores based on dietary parameters that are likely to increase or decrease inflammatory biomarkers (31). Higher positive DII scores reflect pro-inflammatory diets that contribute to systemic inflammation and more negative DII scores reflect anti-inflammatory diets that may protect against systemic inflammation. Chronic systemic inflammation is a significant risk factor for diabetes, obesity, cardiovascular disease, several types of cancer, and numerous other non-communicable disorders that impact healthspan adversely (47). There is compelling evidence that DII scores are clinically relevant and are associated with important health outcomes (48). An analysis of 12 cohort studies that included 18,566 participants and 188,891 person-years of follow-up in the Seguimiento Universidad de Navarra (SUN) study, 6,790 participants with 30,233 person-years of follow-up in the PREvencion con DIeta MEDiterránea (PREDIMED) trial, plus 10 other cohort studies revealed that individuals with the most pro-inflammatory diets based on the highest DII score category had a 23% higher risk of all-cause mortality compared to those with the least inflammatory diets based on the lowest DII score category (49). A meta-analysis of 24 studies found strikingly higher cancer prevalence, cancer incidence, and cancer mortality among individuals in the highest DII categories compared to the lowest DII categories (50). In a study of 34,547 U.S. adults in the NHANES 2005–2018 cycles, 57% of participants had DII scores that reflected pro-inflammatory dietary patterns (51). In the NHANES sample, the weighted mean DII score was 0.44 (SE 0.026). In comparison, in our CALERIE^™ 2 trial, DII values at baseline in the CR and AL groups were −0.28 and 0.04, respectively, and the CR group demonstrated improvements in the DII score throughout the 2 y CR intervention, with an average 2 y value of −1.35. The observed improvements in dietary inflammatory properties in the CR group may have contributed to one or more of the favorable results observed (42–46). As reported previously, DII scores correlated positively with the inflammatory biomarker C-reactive protein at months 12 and 24 in the CALERIE^™ 2 trial (29).

The AL control group, which did not receive nutrition education and was not calorie restricted, on average, based on objective assessments of energy intake, did not exhibit improvements in any of the three metrics of diet quality. The lack of changes in the AL group and the favorable changes in diet quality observed in the CR group further highlight the unique and positive influence of the comprehensive CR intervention.

Nutrient adequacy is important at all ages and life stages for optimal growth, development, physical function, cognitive function, resilience, healthspan, and disease prevention. During long-term CR interventions, avoiding nutrient deficiencies and minimizing losses of bone mineral content and muscle mass are considered vital for maintaining health and avoiding osteoporotic fractures and frailty. Therefore, all participants were provided with a multi-vitamin and mineral supplement to ensure adequacy of key micronutrients. Our current analysis of dietary nutrient adequacy included only dietary sources (without supplements) and was based on sex- and age-specific EAR and AI values. We observed that adequacy varied by specific nutrients, consistent with population data. Specifically, fiber, vitamins D, E, and K, and the minerals calcium, magnesium, and potassium were consistently low at baseline. Other healthy adult populations similarly report inadequacy in these nutrients (52). In addition, our findings are consistent with current nutrients of public health concern noted in the Dietary Guidelines for Americans (38).

In addition to nutritional adequacy not being compromised in the CR group, it is noteworthy that without including the nutrients in the dietary supplements that we provided to participants, more than 90% of CR participants achieved 100% of the EAR or AI for several key nutrients, including protein and a range of vitamins and minerals. This suggests that, despite a moderate reduction in overall calorie intake, participants' micronutrient intake was maintained or improved for the majority of essential nutrients. Dietary fiber, vitamin D, calcium, and potassium continued to be low during the 2 y CR intervention and AL control condition, supporting the provision of supplements that contained fiber, vitamin D, and these minerals. The concurrent low levels of calcium intake coupled with persistently low vitamin D intake signify a common and important nutritional concern. Calcium and vitamin D are intricately linked, with vitamin D playing a pivotal role in calcium absorption and utilization. Low vitamin D intake may compromise calcium absorption and subsequently bone quality, bone mineral density, muscle function, and nerve transmission. In the CALERIE^™ trial, we monitored bone mineral density throughout the intervention and had clear stopping rules if a participant’s bone mineral density decreased ≥5% from baseline (24). Comprehensive strategies to address calcium and vitamin D intake, including dietary modifications, supplementation, and increased exposure to sunlight for enhanced vitamin D synthesis, should be considered to prevent dietary insufficiencies, mitigate the risk of bone loss, and promote optimal bone health and overall well-being in individuals practicing long-term CR.

Our study has limitations. First, dietary intake data were not adjusted for potential under-reporting of energy or nutrients. Therefore, the true diet quality and nutrient adequacy scores may be higher than what we present. In support of this, as shown in Supplemental Figure 1 and as reported previously (29), self-reported energy intake based on food records was ~14–16% lower than objectively-determined energy intake based on DLW. Nevertheless, the quality of the self-reported data was relatively high and the potential impact of under-reporting was minimized in this study due to our participants’ extensive training, practice, and measurement tools to estimate portion sizes accurately. Second, our analyses did not include the vitamin and mineral supplements that all participants received. Therefore, the true vitamin and mineral adequacy is higher than what we report from dietary sources alone. Third, the generalizability of our findings may be limited to healthy, relatively young adults without obesity who follow a moderate CR diet. Thus, we cannot extrapolate our results to individuals who may follow a more restricted diet, older individuals, or individuals with obesity. Future studies should explore the interplay of varying levels of CR and different dietary patterns in individuals of varying ages and health statuses on diet quality and nutrient adequacy. A personalized approach will contribute to optimizing the benefits and minimizing potential risks associated with long-term CR.

In conclusion, a 2 y moderate CR intervention, complemented by comprehensive nutrition education and support, demonstrated favorable diet quality results without compromises in nutritional adequacy in healthy adults without obesity. Future research will be important to explore personalized approaches to enhance dietary patterns that optimize diet quality and ensure adequacy of specific nutrients during longer-term CR interventions.

Abbreviations:

AI

adequate intake

AL

ad libitum

CALERIE

Comprehensive Assessment of Long-Term Effects of Reducing Intake of Energy

CR

calorie restriction

DII

dietary inflammatory index

EAR

estimated average requirement

HEI

healthy eating index

Data Availability

CALERIE^™ 2 trial data sets and code book are available on the CALERIE website at http://calerie.duke.edu). Diet quality and nutritional adequacy data reported in this manuscript are available upon request.