Can We Accurately Predict Adult Height in Pediatric Patients Who Undergo Treatment for Sarcoma?

BACKGROUND

Sarcoma and its treatment has the potential to limit adult height attainment in skeletally immature patients. However, evidence on the extent of this limitation in sarcoma specifically is mixed, and existing height prediction tools such as the Paley multiplier method have proven unreliable in this setting. As such, orthopaedic surgeons are left with the challenge of counseling patients and their families on expected height deficits without an adequate understanding of the extent of these deficits. Equally important, these surgeons must also understand the amount of skeletal growth remaining during the presurgical planning process for limb reconstruction to adequately grasp a patient's risk of subsequent limb length discrepancy.

QUESTIONS/PURPOSES: (1) To what extent does pediatric sarcoma and its treatment limit adult height attainment? (2) Using retrospective data on pediatric patients with sarcoma, can we create a height prediction model that yields more accurate estimates of expected adult height than the Paley multiplier method?

METHODS

For this retrospective pilot study, 223 pediatric patients with sarcoma from a single pediatric sarcoma center between 1976 and June 2022 were identified using diagnostic codes. Inclusion criteria were completion of chemotherapy before skeletal maturity, survival to maturity, and complete height data (that is, height at diagnosis and at skeletal maturity). Of the 223 patients identified as potentially eligible, 56 met inclusion criteria. The remaining 167 patients were excluded on the basis of not receiving chemotherapy (9% [15 of 167]), receiving chemotherapy after skeletal maturity was reached (20% [33 of 167]), not surviving to skeletal maturity (19% [31 of 167]), not reaching skeletal maturity at the time of chart review (43% [72 of 167]), having insufficient treatment data available for analysis (7% [11 of 167]), or being lost to follow-up (that is, no further clinic visits where height at skeletal maturity was recorded) (3% [5 of 167]). Data collection encompassed cancer type; age, height, and weight at diagnosis and maturity; and treatment characteristics. A total of 43% (24 of 56) were female and 57% (32 of 56) were male. Among included patients, 70% (39 of 56) had primary bone tumors, of which 64% (25 of 39) involved lower extremity. Diagnoses of osteosarcoma (41% [23 of 56]) and Ewing sarcoma (36% [20 of 56]) predominated. Doxorubicin (82% [46 of 56]) and cyclophosphamide (61% [34 of 56]) were the most common chemotherapeutics; the mean ± SD treatment duration was 76 ± 88 weeks. Female patients were diagnosed at a mean age of 11 ± 4 years, reaching skeletal maturity at 16 ± 1 years. Male patients were diagnosed at a mean age of 14 ± 3 years, reaching skeletal maturity at 17 ± 1 years. We compared CDC z-scores, which quantify patient height relative to the population mean using SD and percentiles, at diagnosis and maturity to track growth before and after sarcoma treatment, thus quantifying our cohort's growth trajectories relative to the population mean. The Paley multiplier method is a quick and easy-to-use limb length prediction tool that utilizes validated age- and sex-specific multipliers. Its predictions based on height at diagnosis were compared with final attained heights. A novel height prediction model accounting for chemotherapy was derived via multiple regression using two-thirds of the cohort. The remaining one-third was used to test accuracy of the model.

RESULTS

Skeletally immature patients diagnosed with and treated for sarcoma were found to achieve shorter heights at maturity than predicted by the Paley method and CDC growth curves. The Paley method overpredicted adult height by a mean ± SD of 4.3 ± 5.9 cm in female patients (95% confidence interval 1.8 to 6.8; p < 0.001) and 4.9 ± 6.1 cm in male patients (95% CI 2.6 to 7.1; p < 0.001). Median (range) CDC z-scores at maturity (0.47 [-1.7 to 3.2]) were lower than at diagnosis (0.73 [-1.7 to 2.6]) (median difference -0.26; p < 0.001). Patients diagnosed at younger ages were more likely to have larger discrepancies in both CDC z-scores at maturity and in predicted heights. Our novel height prediction model yielded predictions most similar to actual heights attained by incorporating sex, age at first chemotherapy, and height at diagnosis into the following equation (result given in centimeters): 116 + 12 (if male) + 0.6 × (height at diagnosis in cm) - 3.2 × (age at first chemotherapy in years).

CONCLUSION

Our novel height prediction model more accurately accounts for growth limitations imposed by pediatric sarcoma treatment compared with the Paley method and CDC z-scores. With our model, surgeons are now better equipped to plan complex surgical reconstruction of lower extremity tumors in growing children because of a better grasp of eventual adult height and, relatedly, risk for limb length discrepancies. However, to identify any factors that may influence the model's accuracy, larger studies with more diverse cohorts are needed.Level of Evidence Level IV, therapeutic study.