Seven healthy, male, nonsmoking, test subjects participated in the “nutritional countermeasures” (NUC) bed rest study. Female subject were excluded due to a higher risk of venous trombosis during immobilization (Bleeker et al., 2004). This study was funded by the German Aerospace Center (Deutsches Zentrum für Luft-und Raumfahrt, DLR) and the European Space Agency. The subjects had mean age: 27.3 ± 3.4 years, mean body mass: 78.3 ± 6.6 kg, mean body mass index (BMI): 24.1 ± 1.9 kg/m², and VO₂max: 39.5 ± 5.4 ml/kg body mass/min. Approval for the study was obtained from the Ethical Committee of the “Ärztekammer Nordrhein” in Düsseldorf, Germany, and was conducted in accordance with the Declaration of Helsinki. Volunteers gave their written consent after receiving detailed information about the study protocol and the resulting risks. An extensive medical record for each subject was taken, while a routine medical examination and laboratory analyses were done to exclude chronic diseases. None of the subjects received any medication.

Subjects were confined to the bed rest facility of the DLR Institute of Aerospace Medicine, Cologne, Germany for two cross-over designed study campaigns, each consisting of 7 days of pre-bed rest (BDC-7 to BDC-1), 21-days of strict six-degree head down tilt bed rest (HDT1 to HDT21) and 6 days of post bed rest recovery (R + 0 to R + 5; phase R1). Additional follow-up visits took place after 14 (R + 14) and 28 days (R + 28) of bed rest (phase R2). During the 21-day bed rest period, subjects received potassium bicarbonate (3 × 30 mmol day^-1) as a nutritional supplement aimed at preventing an increased bed rest induced bone resorption as the primary outcome of the study. We focussed upon measuring Ca isotopes in the second part of the overall HDTBR study, which involved four test subjects (B, D, F, and H) with dietary supplementation and three without (A, C, and G) during the bed rest period.

During the HDT period, the subjects stayed in bed 24 h a day and all activities (including showering, eating, and weighing) were performed in head-down-tilted-position, not allowing the subjects to raise their heads above 30° from normal. During the non-bed rest phases (adaptation and recovery), the subjects were allowed to walk around in the ward and to do some exercise tests.

Each volunteer received an individually tailored, weight-maintaining diet. The resting metabolic rate was computed by resting indirect calorimetry on the first day in the ward (Deltatrac II MBH 200 Metabolic Monitor). Energy intake was calculated from the resting metabolic rate and an assumed physical activity level of 1.4 for the non-bed rest phases and 1.1 for the bed rest phases. The total energy intake was augmented by 10% to account for diet-induced thermogenesis. Protein intake was 1.2 g/kg BM/day, 30% of calories were supplied as fat, and the remaining energy calories were from carbohydrates. Daily intakes of calcium (1000–1150 mg/day), fluid (50 ml/kg BM/day), sodium (2.8 mmol/kg BW/day) and potassium (100 mmol/day) were also predefined and strictly controlled during the stationary study period. Minerals and vitamins matched the minimum value for the dietary reference intake (DRI, National Academy of Sciences, Institute of Medicine; Food and Nutrition Board, United States). Because subjects were not exposed to the sun, the dietary intake of vitamin D3 was supplemented (1000 IU/day). No caffeine, or alcohol consumption was allowed. All foods were weighed exactly, and each volunteer was asked to consume the complete meal.

Twenty-four-hour urine collection was done during the whole stationary phase period. Continuous urine was collected daily on a void-by-void basis from 7:00 (±15 min) to 7:00 (±15 min) on the following day. Single voids were stored under darkened and cooled conditions and subsequently pooled to 24 h volumes. Several aliquots of the 24-h samples were stored in the freezer at -20°C for later analysis. Samples on day BDC-6, -2; HDT 2, 10, 21; R + 0, + 2, + 5, + 14, + 28 were processed for later analysis for Ca isotopes. After acidifying the urine sample with 2 mol/l HCl, the samples were aliquoted and frozen at -20°C.

Twenty-four-hour urinary bone resorption marker N-terminal telopeptide (UNTX) was determined with a commercially available assay (UNTX: Osteomark^®, Wampole Laboratories, United States) in the in-house laboratory of the Institute of Aerospace Medicine (Cologne, Germany). Intra- and interassay-variation were as follows: Intra: NTX 1.3%; Inter: UNTX 4.9%.

Urine containing 25 μg of Ca was pipetted into PFA beakers and 1 ml high-purity concentrated HNO₃ and 200 μl of H₂O₂ (Suprapure^®, Merck KGaA, Darmstadt) were added. This mixture was placed on a hot plate at 120°C for more than 12 h. After chemical digestion, a small aliquot of the solution containing 2 μg of Ca was taken and an appropriate amount of ⁴³Ca-⁴⁶Ca double-spike tracer solution was added (Galer, 2007). Calcium was separated from the rest of the sample using standard cation exchange techniques utilizing quartz glass chromatographic columns containing ∼0.7 mL of Bio-Rad AG 50W-X8 (200–400 mesh) resin and 2.0 M HCl as eluent. The calcium fraction was dried, and a portion of about 1 μg Ca dissolved in 5 wt.% HNO₃ and loaded onto pre-degassed single or double Re filaments. For single filaments, a Ta₂O₅-based activator was loaded as well, as described in Heuser et al. (2002). Loading onto double filaments was different: first, 1 μl of 0.01 M H₃PO₄ was placed onto the filament and dried down; then the sample solution was pipetted onto the filament and dried down; finally, the filament was raised to a current of 2.0 A for 30 s.

Calcium isotopic compositions were measured at MPIC by thermal ionization mass spectrometry (TIMS, Thermo Fisher Scientific, Triton) in static multi-collection mode. All raw data were corrected for instrumental bias using the double-spike reduction algorithm assuming the exponential fractionation law (Galer, 2007).

All results are reported using the common δ notation for stable isotope variations (Eisenhauer et al., 2004; Coplen, 2011). δ^44/42Ca values (in per mil, ‰) were expressed according to:

Here, the reference material used for “zero δ” is NIST SRM-915a, in common with other studies. Samples enriched in light Ca isotopes (isotopically “light”) compared to the reference material have negative δ while isotopically “heavy” samples (heavy isotope enriched) have positive δ-values. Note, that δ^44/42Ca are a factor of ∼2 smaller than δ^44/40Ca values, a notation often reported in previous work. δ^44/42Ca can be converted to δ^44/40Ca by multiplying by 2.05 (see Heuser et al., 2016).

Statistical analysis and bivariate correlations were performed using the R programming language (The R Foundation for Statistical Computing)¹. Means and single standard errors (1SE) are given throughout the text. P-values for correlations were calculated using a t-test approach and are regarded as significant when P < 0.05.

The primary aim of the study is to examine the temporal evolution of -6° head-down-tilt bed rest on Ca isotopes in urine. To evaluate trends with time we used a linear mixed effect (LME) statistical model. The statistical tests were performed with time as the fixed effect and the subjects as the random effect. Time was initially coded as an ordering factor, and the two BDC measurements from days BDC-6 and BDC-2 were lumped together into one BDC factor level. Homoscedasticity (homogeneity of variance) and normal distribution of residuals were evaluated using residual plots and quantile-quantile plots. Where significant effects of time were revealed by ANOVA, they were further studied by a-priori contrasts of the treatment type. Data are reported as best linear unbiased predictors (BLUP) and standard error (SE). The level of statistical significance was set to 0.05.

Boxplots of the δ^44/42Ca_urine for each of the subjects A to H (Figure 1) show that the urinary Ca isotopic compositions during pre-bed rest BDC span a wide range and are different for most of the subjects. The maximum difference of about 0.8‰ is observed between subjects A and C. Such huge inter-individual Ca isotope differences have been observed previously for urine and blood samples from other bed rest studies (Skulan et al., 2007; Morgan et al., 2012; Channon et al., 2015) as well as in archeological bones (Reynard et al., 2010). Figure 2 shows a comparison of the observed inter-individual ranges in δ^44/42Ca_urine and δ^44/42Ca_blood. The span of δ^44/42Ca_urine in our dataset is the same order of magnitude as that already reported in the literature (Skulan et al., 2007; Morgan et al., 2012).

In their pilot study, Heuser and Eisenhauer (2010) analyzed δ^44/40Ca_urine of a 63-year-old woman suffering from osteoporosis and a 4-year-old healthy boy. A difference of 1.1‰ in δ^44/40Ca (0.55‰ in δ^44/42Ca) was observed between their average urinary Ca isotopic compositions. This offset was attributed to differences in the net bone balance of the two subjects. While the woman had a negative bone balance (loss > gain) the opposite was true for the boy. It can be assumed that the subjects A to H of this bed rest study are not very different in their pre-bed rest net bone balance. Thus, it is unlikely that net bone balance could be the main cause of the inter-individual differences in δ^44/42Ca observed in Figure 1.

The principal finding of this study is that δ^44/42Ca is systematically lowered during bed rest. However, this lowering is occurring with a time delay of 10 to 21 days. After re-ambulation, δ^44/42Ca returns back to baseline values again, after a time delay of 6 to 14 days. Similar time delays have been observed for biochemical markers of bone resorption at the onset and termination of bed rest (Armbrecht et al., 2010) Furthermore, bone mineral content of the tibia decreases for another 10 to 20 days after re-ambulation following bed rest (Rittweger et al., 2010) and after unilateral limb suspension (Rittweger et al., 2006).

Based upon the Ca isotope dataset of our study, we can establish that time delays in the bone’s response to disuse and re-ambulation are present on the levels of: (1) the organic matrix turn-over, (2) calcium trafficking between body fluids and bone, and (3) the anatomical bone structure. As hypothesized by Rittweger et al. (2016), we propose that differentiation and migration of osteoclast precursors and of osteoblasts provide the most likely explanation for the time delays observed. If so, this would imply that determination of bone precursor cells may be a more potent driver of bone’s mechanical adaptation than modulation of existing bone cells.

One also needs to consider that differences in dietary Ca isotopic composition (δ^44/42Ca_diet) could be the origin of the observed differences in the δ^44/42Ca_urine. But since all test persons received a similar diet, inter-subject differences caused by δ^44/42Ca_diet should decrease during the course of our study. However, this is not observed (cf. Table 1); rather, the inter-individual differences stay nearly constant during the whole study period. In addition, it has been shown that the Ca isotopic composition of a diet that is dominated by dairy products (Chu et al., 2006; Heuser and Eisenhauer, 2010; Heuser, 2016) does not vary as much as the 0.8‰ range observed in our study. In summary, δ^44/42Ca_diet probably accounts for only a minor portion of the observed inter-individual differences of δ^44/42Ca_urine.

The observed differences in average pre-bed rest δ^44/42Ca_urine are also not correlated to age (r² = 0.11, P = 0.858), weight (r² = 0.21, P = 0.295), body height (r² = 0.11, P = 0.467) or the BMI (r² = 0.26, P = 0.239) of the subjects (Figure 3).

It is possible to calculate δ^44/42Ca_urine using a Rayleigh-type fractionation model (Heuser and Eisenhauer, 2010):

where α is the fractionation factor between primary and secondary urine, δ^44/42Ca_blood is the input Ca isotopic composition of the blood and f_excreted is the percentage of Ca not reabsorbed in the kidneys, that is collected in the bladder and then excreted. The fractionation factor α is a measure for the amount of isotopic fractionation in an exchange reaction. Assuming the same fractionation factor α between subjects it can be seen from Equation (2) that differences in f_excreted between subjects have a major influence on δ^44/42Ca_urine and should also be reflected in the amount of Ca excreted via urine. In particular, the more Ca that is reabsorbed from primary urine, the less total Ca that will be excreted, and the higher the δ^44/42Ca_urine becomes. This can be tested since differences in the fraction of Ca being reabsorbed in the kidneys between subjects should be reflected in the amount of Ca excreted via the urine.

Indeed, a significant correlation (r² = 0.93, P < 0.001) does exist between δ^44/42Ca_urine and total excreted Ca (mmol/24 h), as can be seen in Figure 4. This correlation suggests that variable individual renal Ca reabsorption is the main cause of the observed inter-individual differences in δ^44/42Ca_urine.

Furthermore, the amount of renal Ca reabsorption is not directly linked to dietary Ca input. The dietary Ca intake between all test persons varied by a factor of 1.2, being only slightly lower than the amounts of individual excreted Ca (factors of 1.3 to 2.3). The differences in the excreted urinary Ca between the subjects varied by, on average, a factor of 4.3 ± 1.3 (1SD). This pattern of higher differences between subjects compared to differences over the course of study for an individual can also be observed in the δ^44/42Ca_urine dataset, which shows smaller intra-individual variation than inter-individual differences.

A change in the fractionation factor during renal Ca reabsorption cannot be excluded. Such a fractionation factor change might plausibly be linked to the amount of cells being involved during transcellular and paracellular transport of Ca in the kidneys, which is ultimately controlled by different hormones (e.g., Mundy and Guise, 1999; Hoenderop et al., 2005; Peacock, 2010). Such a scenario might also change the fraction of Ca reabsorbed, with both effects combining and amplifying the net extent of Ca isotope fractionation in the urine.

In general, δ^44/42Ca_urine decreases during the bed rest phase (Figure 5). This change in δ^44/42Ca_urine is fully in accord with isotopically “light” Ca being released from bones during bone resorption and, indeed, the changes during bed rest in our study match those reported in the literature (Skulan et al., 2007; Morgan et al., 2012; Channon et al., 2015; see Figure 6). We found a change, on average, of about -0.15‰ over the time of bed rest, while Skulan et al. (2007) reported a change of -0.48‰ (control group, 17 weeks of bed rest) and, similarly, Morgan et al. (2012) found -0.2‰ (4 weeks of bed rest). The differences in δ^44/42Ca_urine shifts during bed rest periods observed between studies can most likely be attributed to the different length of the bed rest phase in each case.

Our Ca isotope dataset shows a small difference between the group of subjects with and without supplement of potassium bicarbonate (KHCO₃), although this is statistically not significant (Figure 6). Specifically, the mean change of δ^44/42Ca_urine from baseline is -0.21‰ for subjects without KHCO₃ supplement, while it is -0.11‰ for those who took the supplement.

Up to now, changes in δ^44/42Ca_urine have been interpreted to be the consequence of bone resorption only. Support for this explanation comes from (1) the data of Channon et al. (2015) who, like in the present study, found a decrease in δ^44/42Ca_blood during bed rest, and (2) from Skulan et al. (2007) who reported a correlation between δ^44/42Ca_urine and changes in conventional serum biomarkers for bone resorption. Other physiological effects during bed rest that might lead to changes in δ^44/42Ca_urine have not yet been fully investigated, however.

For example, assuming a constant amount of Ca transported from the intestine into blood along with prolonged bone resorption during bed rest, the Ca concentration in blood must increase and will be regulated by an increase in overall Ca excretion. But, in order to increase Ca excretion, the fraction of Ca that is reabsorbed in the kidneys (f) must decrease. In terms of Ca fractionation this will result in lower δ^44/42Ca_urine as increased amounts of Ca are being excreted (cf. Figure 4). A decrease in the fraction of Ca being reabsorbed from 98 to 97% would cause a lowering of δ^44/42Ca_urine by about -0.1‰. Thus, the shift in δ^44/42Ca_urine during the bed rest phase can be explained by either the addition of isotopically “light” Ca from bones to the blood or, alternatively, by a change in the renal Ca excretion factor (f) or, more probably, the combined effect of both.

The observed inter-individual differences in δ^44/42Ca_urine and corresponding δ^44/42Ca_blood from the studies of Morgan et al. (2012) and Channon et al. (2015), respectively, support a changing excretion factor (f) during bed rest. While the inter-individual differences (max δ^44/42Ca–min δ^44/42Ca) are about 0.85‰ for δ^44/42Ca_urine at the end of the bed rest phase, the reported shift in δ^44/42Ca_blood is smaller (0.46‰).

On the other hand, an increase in the concentration of free Ca in the blood triggers a serum parathyroid hormone (PTH) concentration decrease (Mundy and Guise, 1999). Consequently, the decrease of PTH concentration then results in a decrease in intestinal Ca absorption and a lowered renal Ca reabsorption (Mundy and Guise, 1999).

Serum NTX (n-telopeptide cross-links) is a frequently used biochemical marker for bone resorption (e.g., Iba et al., 2008). Skulan et al. (2007) reported changes in NTX alongside δ^44/42Ca_urine in their samples. At first glance their data suggests a correlation between NTX and δ^44/42Ca_urine since mean δ^44/42Ca_urine decreases during bed rest as mean NTX increases. This qualitative correlation suggests that changes in δ^44/42Ca_urine during bed rest are linked to increased bone resorption. Such a qualitative correlation between δ^44/42Ca_urine and changes in NTX is also seen in our dataset. Relative to pre-bed rest values, the mean change in δ^44/42Ca_urine post-bed rest is -0.10‰ while the corresponding change in NTX is 393 (nmol/d). But a closer look at the pooled data shows that there is in fact no statistically significant correlation between δ^44/42Ca_urine and NTX (Figure 7; P > 0.99), nor for any of the subjects individually (cf. Table 2).

Based on our data and those from previous studies, bone loss cannot be detected unambiguously using δ^44/42Ca_urine if the (steady state) pre-bone-loss baseline δ^44/42Ca_urine of the patient is unknown. This hampers the easy application of δ^44/42Ca_urine as a diagnostic tool for detecting disease related to bone loss, such as osteoporosis. Here, the main problem seem to be in the unknown amount of Ca being reabsorbed in the kidneys for each patient. But on the other hand, using δ^44/42Ca_urine it should still be possible to monitor the success of an on-going osteoporosis therapy aimed at reducing bone resorption.

In order to overcome this problem we propose a new baseline approach intended to reduce the unknown renal Ca absorption effect on δ^44/42Ca_urine. In a log-log plot of ⁴⁴Ca/⁴²Ca_urine versus the amount of excreted Ca per day (Figure 8) two different negatively sloped trends can be seen, for the adaption (BDC) and bed rest (R + 0, R + 2, R + 5) phase. The BDC samples follow a linear trend (r² = 0.780, P < 0.001) with a slope of -5.38 × 10^-4 and a y-axis intercept of 1.177, while the bed rest samples linear trend (r² = 0.790, P < 0.001) has a slope of -5.43 × 10^-4 and y-axis intercept of 1.176. Although both trends are not strictly parallel to one another, they remain basically parallel within the range of excreted Ca measured. The offset between the two trends seen in Figure 8 indicates that the Ca isotopic composition for any given Ca excretion rate is always isotopically “lighter” at the end of the bed rest relative to the BDC baseline.

Thus, we suggest that a combined measurement of daily Ca excretion rate as well as δ^44/42Ca_urine may be sufficient to implicate patient bone resorption, if the data lie below the BDC trend when plotted in Figure 8. For the time being, this inference must be considered tentative, pending better definition of the BDC line in Figure 8 from future studies; this can be done relatively easily, however, since it just involves measurement of these two parameters in healthy individuals whose Ca balance is at steady state.

We used the R1 data here, as they should be the most affected by the bed rest phase. Based on Figure 6 it seems justified to assume that the shift from the baseline trend would be even more pronounced during longer periods of bed rest. Figure 6 also shows a small but not significant difference between the control group (C) and the supplemented group (S). While the control group defines a linear trend shifted parallel toward lower δ^44/42Ca, the supplemented group defines a trend with a lower slope and higher y-axis intercept. Currently, we do not have a good explanation for this – it might be related to the supplement of KHCO₃ but may also be simply a statistical artifact due to the limited number of subjects in each group (n_C = 3, n_S = 4).

One should keep in mind that the data currently available, here and elsewhere, are far too few in number to define a statistically meaningful BDC δ^44/42Ca_urine versus excreted Ca rate baseline. Nevertheless, it does reveal a promising way for canceling out the patient-specific effects of renal Ca reabsorption on measured δ^44/42Ca_urine. In order to assess whether such a common BDC baseline approach actually works, many more data will be needed. Furthermore, the influence of dietary Ca isotopic composition on δ^44/42Ca_urine and other potential fractionation mechanisms in the body have to be better understood, and represent significant gaps in our present knowledge that must be addressed in future studies.