Scoliosis, Cement, and a Crooked Auto-ROI
Your next patient is a 61-year-old man, 178 cm and 118 kg, monitored after 2 years of androgen-deprivation therapy for prostate cancer. He has a 25-degree right thoracic scoliosis, a T12 vertebroplasty with cement, and bilateral hip osteoarthritis severe enough that he cannot lie flat with his legs internally rotated. His right femoral neck was pinned 10 years ago after a fall. The prior scan, done at another clinic on a different manufacturer's densitometer, reported L1-L4 BMD 1.020 g/cm2 and total hip 0.880 g/cm2. Today's spine acquisition is done with him positioned as comfortably as he can manage. On the analysis screen, the T12 cement sits at the upper edge of the region box, and the auto-analysis has drawn the L4 inferior border through the L4-L5 disc space.
1. How do you position and analyze the spine given the scoliosis, the cement, and the misplaced region border?
Read the worked answer
- Positioning: supine, spine midline on the table, arms flat at the sides (not overhead - overhead arms alter soft-tissue subtraction). Use the leg block only to the extent he tolerates it; if forced internal rotation causes pain or movement, document the compromise rather than producing a motion artifact.
- Numbering: the cement at T12 is a gift - use it as an internal landmark to confirm level labeling, starting from T12/L1. Scoliosis and rotation can shift the soft-tissue baseline; check that each vertebral box actually encloses the vertebral body and that the intervertebral disc spaces sit between boxes.
- Correct the region: redraw the L4 inferior border through the vertebral endplate, not the disc space. Auto-analysis fails more often in scoliotic spines.
- Exclusions: the cemented T12 is outside a standard L1-L4 region, so leave it out; also exclude any compressed, fractured, or instrumented vertebra. Do not diagnose or monitor from a single vertebra.
- If fewer than two evaluable vertebrae remain, delete the spine analysis and rely on the hip or forearm, stating why.
2. Which hip and which region would you analyze, and what would you do if neither hip were usable?
Read the worked answer
- The right hip is unusable for BMD. A pinned femoral neck places metal directly in the femoral neck and total hip regions of interest; hardware invalidates the measurement. Do not simply 'work around' it.
- Use the left hip. Osteoarthritis affects the acetabulum and the superomedial joint space and inflates total hip BMD more than femoral neck BMD. Report the femoral neck as the primary value and note that total hip may be artificially elevated by degenerative change.
- If both hips were invalid (for example, bilateral arthroplasty or bilateral hardware), the fallback site is the 33% radius of the non-dominant forearm, clearly labeled as such.
- Be consistent: compare future scans with the same side and the same region. Changing which hip you report turns measurement noise into apparent change.
3. What can you say about change since the outside scan, and how do you word the report?
Read the worked answer
- Different manufacturers cannot be compared numerically without cross-calibration. Systematic offsets of several percent exist between vendors, and they differ by site, so the outside 1.020 g/cm2 and today's number are not on the same ruler.
- Even on one machine, a change must exceed the facility's least significant change before anyone calls it real.
- The segments may also differ: if you have excluded levels or corrected a misdrawn region, you may not be comparing like with like even ignoring the device issue.
- Report language: state the current spine and hip values with their excluded levels and the reason; state plainly that comparison with the outside study is limited by a different manufacturer, no cross-calibration, and differing analyzed segments; recommend the current values as the new baseline for future monitoring on this device; flag the T12 cement and the right hip hardware.
- Do not tell the patient he has 'gained' or 'lost' bone since the last scan; you cannot support either claim from this data.