genetic, hormonal, and nutritional factors contribute to bone metabolism, but the daily mechanical loads experienced by bone also determine bone size and shape (6, 10, 14, 19). When new forces or loads alter the normal daily pattern of bone bending or strain, the bone adapts by increasing formation that, in turn, increases mass, size, and moment of inertia to resist the altered bending. The adaptation is seen with exercise training (11, 13, 27, 28, 31,36-38), external loading (7, 25, 29, 34), and osteotomy (4, 5,20). The early bone changes associated with mechanical loading include increased mRNA expression, osteoblast number, and bone formation within 5 days after loading (3, 9, 12, 23). Others have shown that the adaptive response also includes the suppression of bone resorption (17,37).

If mechanical loads regulate bone size, then an increase in loading should result in only a transient acceleration in bone formation. Bone cell activity should increase until the new structure is sufficient to meet the new strength demands. Once adaptation is complete, cell activity should return to nonexercise levels, and the new structure should be maintained. A permanent acceleration of bone formation would overcompensate for a discreet load change. The pattern should resemble muscle training, in which muscle mass and strength plateau until the training level is increased.

Some studies have indirectly shown the waning of bone formation with time (29, 36). The only study that identified a cessation of formation also reported that bone size was reduced from 6 to 14 wk of loading and suggested that the bone consolidated as a final adaptation to loading (34). However, that conclusion was based on cross-sectional data; a method for consolidation was not documented, and osteoclast activation was not observed. Loss of bone area with maturation or time is not commonly associated with growth. A clear understanding of bone cellular and structural adaptations to increased stress is important for the interpretation of serum and bone measures in athlete cross-sectional and exercise-intervention studies.

The purpose of this study was to examine the time course for bone adaptation to controlled, external, mechanical loading in the rat tibia four-point bending device. Our hypothesis was that consistent long-term load application causes a transient increase in bone formation and an incremental gain in periosteal bone that is maintained with continued loading. To test this hypothesis, rats were loaded for up to 18 wk, and tibial formation surface (FS), formation rate, and new bone thickness (newBTh) were compared with contralateral nonloaded tibial values.

MATERIALS AND METHODS

The effects of long-term, external mechanical loading were studied in the tibiae of female, retired Sprague-Dawley breeder rats (SASCO, Omaha, NE) (6 mo and 282 ± 30 g). The rats were individually housed in wire cages (20 × 24 × 18 cm), and food and water were available ad libitum. All procedures were approved by the University's Institutional Animal Care and Use Committee. Forty-seven rats were randomized by body weight to four external loading (L) groups. One group was loaded Monday-Wednesday-Friday (M-W-F) for 6 wk (n = 11), whereas three groups were loaded every other day for 6 wk (n = 11), 12 wk (n = 12), or 18 wk (n= 13). Fourteen rats acquired within a week of the loaded rats were maintained as an 18-wk control group. All rats continued unrestricted, weight-bearing activity while in their cages.

An external mechanical load was applied to the lower right leg with a rat tibia four-point bending device (1, 26). The upper pads in the loading device were 10.5 mm apart and centered between the lower pads, which were 22 mm apart. The maximal bending region was between 3.5 and 14 ± 0.5 mm, proximal to the tibia fibular junction (26). Bending moments of 92 N ⋅ mm (32 N) were applied at 2 Hz for 36 cycles to create 1,200 microstrains (με) in compression on the lateral tibial surface and 1,700 με in tension on the medial tibial surface. The left leg was not loaded and served as a contralateral control.

All loaded rats were given an intraperitoneal tetracycline (25 mg/kg) injection 1 day before the experiment to label preexisting bone-forming surfaces. All rats were given an intraperitoneal calcein (Sigma Chemical, St. Louis, 8 mg/kg) injection 11 and 3 days before death. At 6, 12, and 18 wk, rats were anesthetized and killed by intracardiac injection (0.1 ml, FatalPlus, Vortech Pharmaceuticals, Dearborn, MI). The left and right tibiae were then collected, fixed in 70% ethanol, block-stained in Villanueva stain (35), dehydrated in ethanol and acetone, and embedded in methyl methacrylate (2). Cross sections of 120-μm thickness were cut 4–8 mm proximal to the tibia-fibular junction with a low-speed diamond wheel saw (model 2680, South Bay Technology, Temple City, CA) and ground to 80-μm thickness for mounting. Two sections from each tibia were blind coded and analyzed, and their data were averaged.

Periosteal and endocortical surfaces were digitally traced with a microscope, camera lucida, graphics pad, and the BIOQUANT semiautomated image-analysis system II (R&M Biometrics, Nashville, TN). Measurements included cortical area, woven bone area, periosteal and endocortical perimeters, double and single calcein-labeled perimeters, tetracycline-labeled perimeter, and woven bone perimeter. Woven bone was identified by irregular diffuse labeling patterns. Interlabel thickness (IrLTh) was directly measured at equal intervals between all double calcein labels. newBTh was measured between the initial tetracycline label and the current periosteal surface. The length of each unique type of surface was reported as a percent of the total bone surface (BS): single calcein-labeled surface (sLS/BS), double calcein-labeled surface (dLS/BS), and woven bone bearing surface (WoS/BS). Formation surface was the sum of the three unique forming surfaces {FS/BS = [sLS/(2 + dLS + WoS)/BS]}. Mineral apposition rate (MAR) was calculated at all dLS sites as the interlabel thickness divided by interlabel time (IrLTh/8 days). Surface-based bone formation rate was calculated as (MAR × FS)/BS (21). The average MAR (aveMAR) for the duration of the study was calculated at sights with an initial tetracycline label (aveMAR = newBTh/days on study)

Strain (με) on the lateral tibial surface during four-point bending was calculated from a regression equation based on cross-sectional moment of inertia, beam-bending theory, previous in vivo strain gauge measurements, and mechanical testing (1). Two cross sections from each tibia were traced at ×20, and the moment of inertia about the anterior-posterior axis was computed with Section (Biomechanics Lab, Creighton University) on a VAX Station 2000 computer. Strain was calculated and averaged for two sections of the loaded tibia. Measurement of the final moment of inertia was used to estimate strain after adaptation to loading.

The two 6-wk loading groups (M-W-F and alternate days) were tested with Student's t-test for differences in response to loading. Because there were no differences between these groups, the groups were combined to form a single 6-wk loading group. ANOVA Split-Plot Design (CRUNCH, Crunch Software, Oakland, CA) was applied to the data from all groups. The split plot design analyzed two factors: 1) differences between control and loaded tibia within rats for each group and 2) differences between all treatment groups and the control group. When the between-group effect was significant, the differences were tested with Bonferroni post hoc tests. When a significant interaction occurred (load × week), the loading effect was calculated as loaded − nonloaded, and differences between weeks were tested. Differences were considered significant at P < 0.05 for all tests.

RESULTS

The rats of all five groups were similar in size. Initial body weights averaged 282 ± 30 g and were not different between groups at any time. Although body weight increased within each group from initial to final time points, there were no differences in final body weight (Table 1). Final cross-sectional area and moment of inertia of the tibiae were not different between groups or between loaded and nonloaded tibia within groups (Table 1). Strains during loading, estimated from final moment of inertia, were not different among groups and averaged 1,200 ± 165 (SD) με on the lateral surfaces, 1,680 ± 230 με on the medial periosteal surfaces, and 900 ± 125 με on the endocortical surfaces.

Although the two groups loaded for 6 wk had different daily loading schedules, there were no differences between the groups in labeled surface, FS, MAR, or newBTh. Therefore, these two groups were combined for analysis into a single 6-wk group. Of the 47 rats externally loaded for 6, 12, or 18 wk, one rat from the 12-wk group died due to anesthetic overdose.

Within-group differences due to loading.

Loading effects at each time point were determined by examining differences within groups between the loaded and nonloaded contralateral tibiae. The primary response to external loading was organized, compact periosteal and endocortical bone formation. However, small amounts of woven bone were identified on four loaded tibial periosteum. In those four rats, the woven bone perimeter averaged 9.2 ± 3.2% of the total periosteal length and 3.2 ± 1.8% of the total cortical area. Periosteal resorption or remodeling was not seen in the cortical bone at any time after loading.

At 6 wk, the periosteal surface-loaded tibia had greater (P < 0.0007) FS, MAR, and bone formation rate than the contralateral tibia (Table 2). After 12 wk of loading, FS was still greater in loaded than nonloaded legs; however, MAR and bone formation rates were not different due to loading. There were no differences within groups after 18 wk of loading or in the 18-wk control group. newBTh was greater in the loaded than in the nonloaded legs after 6 wk of loading. The difference between legs was not significant at 12 or at 18 wk. The periosteal aveMAR for the duration of study was greater in the loaded legs at 6 wk but not different at 12 or 18 wk.

On the endocortical surface, the 6- and 12-wk loaded tibiae had significantly greater (P < 0.0005) FS, MAR, and bone formation rates compared with the contralateral tibiae (Table3). After 18 wk of loading, comparison with the contralateral nonloaded tibiae demonstrated that the loaded tibiae MAR was still elevated (P < 0.05). There were no significant differences on the endocortical surfaces within the 18-wk control group. The total newBTh on the endocortical surface was greater in the loaded leg after 6 wk of loading (P < 0.02). There was no difference in newBTh after 12 or 18 wk of loading. The endocortical aveMAR for the duration of study was greater in the loaded legs at 6 wk but not different at 12 or 18 wk.

Differences between groups with time.

The effects of time or aging were examined as differences among groups in formation (collections at 6, 12, and 18 wk) in the nonexternally loaded tibia (Table 2). Periosteal FS and MAR were not different among tibia at 6, 12, or 18 wk. However, bone formation rates averaged over the duration of the study and measured during the final 2 wk were lower in the 18-wk than in the 6-wk group. The total periosteal newBTh was not significantly different among the nonloaded tibia.

Endocortical FS, MAR, and bone formation rates were higher in nonloaded tibia at 6 than at 12 or 18 wk, but there were no differences between the 12- and 18-wk groups (Table 3). The total endocortical newBTh was not different among the nonloaded tibia, but aveMAR was greater at 6 than at 18 wk.

Differences in loading effect with time.

The loading effect (loaded − nonloaded) tended to decrease with time on study. On the periosteal surface, the magnitude of the loading effect was greater at 6 wk than at 18 wk for FS, and bone formation rate was greater at 6 wk than at 12 and 18 wk. The differences between tibiae due to loading did not vary among weeks for MAR, newBTh, or aveMAR. On the endocortical surface, the magnitude of the loading effect for bone formation rate was greater at 6 wk than at 18 wk. There were no differences in the loading effect among weeks for FS, MAR, or newBTh. The differences in periosteal FS, periosteal bone formation rate, endocortical MAR, and endocortical bone formation rate between loaded (right) and nonloaded (left) legs was greater for 6-wk loaded rats than for the nonloaded control rats. The only 12-wk difference was between endocortical MAR in the loaded and nonloaded controls.

DISCUSSION

Bone adapts to external loading at 1,200–1,700 με by transiently increasing formation and increasing periosteal newBTh. The elevated formation seen at 6 wk is consistent with previous 3- to 4-wk loading studies of this model (16), and the lack of response at 18 wk is consistent with a previous loading study (34). However, the present study is unique because it examines cumulative bone gain with loading. In the previous 14-wk study, the area of newly formed bone “consolidated” or diminished in size with adaptation (34). In contrast, this study shows gain in periosteal thickness by week 6 of loading that is maintained through weeks 12 and18 of loading.

Age-related changes in periosteal formation.

The decrease in formation in the loaded legs from week 6 toweek 18 potentially reflects both a decrease in loading effect and an age-related decrease in formation. In nonloaded legs, bone formation rate was lower at week 18 than at week 6 of study for both periosteal (70%) and endocortical (94%) surfaces. Although the growth rate at the proximal tibial growth plate is minimal in the 6-mo-old female Sprague-Dawley rat (18), this study shows that diaphyseal expansion slows but continues with aging.

Time course of adaptation to loading.

We found that, although the initial increase in formation parameters with loading was transient, the bone mass effect was maintained. The loading response for bone formation rate diminished at least 5-fold on the periosteum and 14-fold on the endocortical surface from week 6 to week 18. The endocortical loading response persisted longer than the periosteal response, as MAR was still elevated in the loaded leg at week 18.

The pattern of initial rapid formation with a return to control levels was confirmed by measures of newBTh. The total periosteal bone formed from day 0 until the bones were collected was greater in loaded than in nonloaded legs at 6 wk and tended to increase with time. Although both loaded and nonloaded legs tended to have greater newBTh at weeks 12 and 18 than at week 6, and the loaded legs tended to remain greater than nonloaded legs, the differences were small and not statistically significant. One explanation for the lack of difference between legs at the later times could be that the differential rates of formation between legs, with time, resulted in high intra-animal variation.

Bone adaptation to four-point loading appears to be immediate, with a gradual dissipation of formation and a return to normal control levels. The time course for adaptation to loading at 92 N ⋅ mm (1,200 με) in the present study is similar to that observed in a previous study after loading at 120 N ⋅ mm (2,000 με) (34). Both studies show increased bone formation and mineralization rate at weeks 3,6, and 7, decreasing by week 12, and returning to control levels by weeks 14–18. However, the pattern of response was different between the two studies. At 2,000 με, primarily woven bone was produced, whereas at 1,200 με, a more organized, lamellar-like bone pattern was produced. In the previous study, it was suggested that between weeks 3 and 14 of loading at 2,000 με, porous woven bone became consolidated, and cortical area decreased from 180 to 120% of control values. This is in contrast to our study in which the differences between legs in periosteal new bone area were similar at weeks 6, 12, and 18. Our study differs from the previous in that lower forces were applied to create primarily lamellar bone, and the sample sizes taken at each time point were greater. We observed no evidence of remodeling or resorption on the periosteal surface, and the pattern of new bone appeared similar among groups and between legs and remained consistent with the initial bone. Our results suggest that bone formed as a result of moderate loading is maintained with continued loading and does not consolidate or diminish in size.

The observed transient increase in formation followed by a period of maintenance with continued loading is consistent with previous loading studies. Although not originally presented in the same context, the isolated rooster ulna model showed that, when loaded at 2,000 με for 36–1,800 cycles/day, ulnar bone mineral content increased 40% during the first 4 wk of loading and remained stable from week 4 to week 7 of continued loading (29). After ulna osteotomy in dogs, radial cortical area increased 15–30% within 1–3 mo after surgery (4, 5). Periosteal formation peaked at 4–6 wk, returned to normal by 12 wk, and remained normal at 24 wk after surgery. Exercise studies have also demonstrated transient increases in bone formation rate and gains in bone mineral content (36-38). Similar to the imposed loading studies, rats run on treadmills appeared to have greater stress-related differences in formation after 9 wk of training than at 16 wk (36). Gains in bone mineral content and bone mineral density for both cortical and cancellous bone also appeared greater after 0–9 wk than after 9–16 wk of treadmill running (37, 38). These previous studies indirectly demonstrate an incremental gain in bone area.

Consistent with our model, these other skeletal loading models support two principles of bone adaptation. First, when a unique load with a defined magnitude (i.e., 92 N ⋅ mm, 120 N ⋅ mm, or body wt) is repeatedly applied to bone, adaptation occurs within a discrete period of time, and formation returns to a steady-state level similar to age-matched controls. Second, there is an incremental gain in bone size that is maintained with continued loading. This pattern would support Frost's (15) theory of skeletal structural adaptations to mechanical usage. Frost suggests that bone adapts to a normal range of loads, and, when new forces are outside the adapted window, bone modeling occurs to accommodate these forces.

The strain threshold for initiating modeling in the isolated ulna and rat tibia four-point bending models has been reported to be more than 1,000 με (30, 33). At 6 wk in the present study, the applied loads created estimated periosteal strains above this loading threshold. When analyzed separately, the medial surface (higher, tensile, +1,700 με) tended to have a greater formation response than the lateral surface (lower, compressive, −1,200 με), but the differences were not significant. Compared with estimated strains at 6 wk, the initial strains may have been greater, and strains at 18 wk were ∼8% lower. Strains at 18 wk still appeared to be above the predicted threshold, despite cessation of formation. This apparent conflict with Frost's (15) theory could reflect 1) inaccurate strain prediction after new bone formation with altered geometry or2) adaptation to the unique load distribution. Further studies will be needed to determine actual strain after adaptation and to determine if strain thresholds are dependent on load distribution (unique vs. normal or adapted). In contrast to the periosteum, endocortical formation was elevated through 18 wk of loading, despite strain levels below threshold. These data would suggest that the endocortical response is not dependent solely on strain magnitude but reflects total bone stimulation.

Implications for exercise training.

External loading and surgical models apply loads that create similar strain magnitudes but different strain distributions from exercise. Although specific details of loading (number of repetitions, strains, forces) are not directly applicable from external loading models to exercise, the models can offer insight on bone adaptation to mechanical loading. The present study suggests that an exercise or training program that maintains the same loads and activities for many years (i.e., marathon running) would stimulate formation only during the initial months of training. It has been recognized for some time that, with muscle strength training, periodization of the resistance program is necessary to alter the loading stimuli and maximize results (8, 22,24, 32). It seems reasonable that bone would have a similar yet slightly slower response than muscle, and, once the demands of a load have been meet, increased stimuli would be necessary to cause further cellular activation. Therefore, long-term exercise may maintain bone mass at above normal levels with normal levels of cell activity.

In conclusion, external loading in four-point bending at 1,200–1,700 με stimulated periosteal lamellar bone formation. Similar to other long-term bone studies, periosteal formation is elevated transiently for 6–12 wk and returns to age-matched control levels by week 18 of loading. After 12 wk of loading, bone adaptation reached a steady state, with no differences in the newBTh or MAR. Loading that continued beyond week 12 appeared to only maintain the current bone status. Because bone adapts rapidly to a consistent loading pattern, loads should be increased incrementally with time, perhaps every 6 wk, to continue stimulation of formation. In designing future loading experiments, bone cell activity should be examined during the transient period, and mechanical properties should be examined after adaptation is complete.

FOOTNOTES

• This work was supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases Grant AR-39221, National Research Service Award AR-08144, and a grant from Health Future Foundation.

• The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.