9.8 Lumbar Spine Follow-Up Scans and Rate of Change
Key Takeaways
- The lumbar spine shows the largest and earliest response to therapy of the three DXA sites because it is trabecular-rich with high remodeling surface area.
- Reproducing the baseline study means the same scanner, acquisition mode, positioning aids, vertebral labeling, region-of-interest placement, and vertebral exclusions.
- Vertebrae excluded at baseline must be excluded on every subsequent study, because changing the analyzed vertebral set changes the measured quantity.
- Rate of change is reported from absolute BMD in grams per square centimeter, not from T-scores, which shift with reference database and software changes.
- A new compression fracture, new degenerative change, new hardware, or new calcification between studies can invalidate serial comparison even when technique was reproduced perfectly.
9.8 Lumbar Spine Follow-Up Scans and Rate of Change
Quick Answer: The lumbar spine responds fastest of the three sites, which makes it the preferred site for detecting treatment effect. Valid serial comparison requires reproducing the same scanner, the same acquisition mode, the same positioning and aids, the same vertebral labeling, the same ROI placement, and the same exclusions — and comparing absolute BMD in g/cm² against the facility's least significant change.
Why the Spine Is the Monitoring Site
The lumbar vertebral body is a thin cortical shell around a trabecular core. Trabecular bone presents a large surface-to-volume ratio, remodeling occurs on surfaces, and so the spine registers change before the predominantly cortical sites do.
| Site | Typical first-year response to antiresorptive therapy | Monitoring usefulness |
|---|---|---|
| Lumbar spine (L1–L4) | Roughly 2–4% gain | Best; largest signal relative to precision error |
| Total hip | Roughly 1–2% gain | Good; better precision than femoral neck |
| Femoral neck | Roughly 1–2% gain | Usable; poorer precision |
| One-third radius | Minimal change for years | Poor for monitoring |
Untreated postmenopausal loss runs roughly 1–2% per year, and anabolic agents can produce spine gains of 5–10% over one to two years. Those figures set the expectation against which a measured change is judged.
Reproducing the Baseline Study
The ARRT outline names "reproduce baseline study (acquisition parameters, positioning, ROI placement)" explicitly. At the spine that means seven specific things:
| Element | Requirement |
|---|---|
| Scanner | Same unit. A different unit requires cross-calibration or a new baseline |
| Acquisition mode | Same mode. Changing between fast array, standard, and high-definition alters photon statistics and precision |
| Software version | Ideally the same; if upgraded, re-analyze the baseline on the new version where the manufacturer advises |
| Positioning | Same leg block at the same setting, same centering, same arm placement |
| Scan field | Same superior and inferior limits, so the same anatomy is captured |
| Vertebral labeling | Same convention, verified against T12 ribs and the iliac crest |
| ROI and exclusions | Same intervertebral marker placement, same vertebrae included and excluded |
Most systems allow the baseline analysis to be displayed alongside the follow-up. For routine clinical follow-up, use it — matching the prior analysis is how reproducibility is achieved. The one exception is a precision study, where each scan must be analyzed independently or the precision estimate is falsely optimistic.
The Exclusion Rule for Serial Studies
If L2 was excluded at baseline for a bridging osteophyte, L2 is excluded at follow-up. This holds even if the software's automatic analysis would include it, and even if a different technologist would have made a different original decision.
The reason is arithmetic. L1–L4 BMD is the total BMC of the included vertebrae divided by their total area. Adding or removing a vertebra changes both, and the resulting difference has nothing to do with the patient's bone.
Worked illustration. Baseline L1, L3, L4 (L2 excluded) yields 0.842 g/cm². Follow-up analyzed as L1–L4 including the osteophytic L2 yields 0.881 g/cm². The apparent "gain" of 0.039 g/cm² — about 4.6%, exceeding a typical spine LSC of about 0.033 g/cm² — would be reported as a significant therapeutic response. Nothing happened to the patient. The analysis changed.
This is why "document what you excluded and why" is not clerical advice.
Calculating and Reporting Rate of Change
- Compute the absolute difference in BMD: $\Delta \text{BMD} = \text{Follow-up} - \text{Baseline}$.
- Compare to the facility LSC for the spine, derived from that technologist's precision study as $2.77 \times \text{RMS-SD}$.
- Report significance: significant increase, significant decrease, or no significant change.
- Express rate where clinically useful, as absolute change per year or percent change per year.
Worked example. Baseline spine BMD 0.812 g/cm²; follow-up 24 months later 0.861 g/cm²; facility spine LSC 0.033 g/cm².
- $\Delta \text{BMD} = 0.861 - 0.812 = +0.049\ \text{g/cm}^2$
- $0.049 > 0.033$, so the change is statistically significant at 95% confidence.
- Percent change $= 0.049 \div 0.812 \times 100 = +6.0%$ over two years, roughly $+3.0%$ per year.
Second example. Baseline 0.900; follow-up 0.922; LSC 0.033. $\Delta = +0.022$, which is less than the LSC. Reported as no significant change — stable bone density — even though the arithmetic shows a 2.4% rise. A change below the LSC is not reportable as change.
Why Not T-Scores
T-scores are rounded to one decimal, and a change of 0.005 g/cm² can move the displayed T-score across a rounding boundary while a real loss of 0.025 g/cm² hides within one. Reference database and software updates move T-scores with no physical change at all. Serial comparison uses absolute BMD. Reporting the rate of change in T-score terms is descriptive at best.
Events That Invalidate Comparison
Even with perfect technique, some changes between visits break comparability:
- A new vertebral compression fracture at an included level. The level must now be excluded, which changes the analyzed set — re-analyze the baseline with the same exclusion, or restart the baseline.
- New or progressive degenerative change, particularly new bridging osteophytes or advancing endplate sclerosis. Degenerative change adds bone over time, so an elderly spine can show a rising BMD that is entirely artifactual.
- New surgery or hardware at an included level.
- New aortic calcification.
- A different scanner or manufacturer without cross-calibration.
- A software or reference database change affecting the analysis algorithm.
- Markedly different body composition — large weight change alters the soft-tissue baseline.
In each case the correct action is to identify and report the limitation, not to produce a number and let the reader assume comparability.
Timing of Follow-Up Studies
- Typical interval: one to two years after starting or changing therapy, then less often once stability is established.
- Shorter intervals (6–12 months) are reserved for rapid-loss states: high-dose glucocorticoid therapy, transplantation, and acute immobilization.
- Very short intervals (3–6 months) are rarely useful, because expected biological change over that period is smaller than the LSC, guaranteeing an uninterpretable "no significant change" result while adding dose.
- Medicare frequency generally permits a covered study once at least 23 months have passed, with the named exceptions for long-term glucocorticoid therapy and confirmatory baselines.
At baseline, L2 was excluded for bridging osteophytes and the L1, L3, L4 result was 0.842 g/cm-squared. The follow-up was analyzed as L1 through L4, giving 0.881. The facility spine LSC is 0.033. How should this be handled?
Baseline spine BMD is 0.900 and the follow-up is 0.922 g/cm-squared, with a facility spine LSC of 0.033. How is this reported?
Why is the lumbar spine preferred over the one-third radius for monitoring response to therapy?