ARCH Subgroup Analyses: Who Responded Most and Least to Romosozumab?

At a glance

| Parameter | Detail | |-----------|--------| | N | 4,093 postmenopausal women | | Intervention | Romosozumab 210 mg SC monthly x 12 months, then alendronate 70 mg weekly | | Comparator | Alendronate 70 mg weekly x 12 months, then alendronate 70 mg weekly | | Duration | 24 months (primary analysis) | | Primary endpoint | New vertebral fracture at 24 months | | Key result | 48% relative risk reduction vs alendronate (6.2% vs 11.9%; p <0.001) |

Trial Design Relevant to Subgroup Interpretation

ARCH enrolled postmenopausal women aged 55 to 90 with osteoporosis and a fragility fracture history. Unlike the earlier FRAME trial (which used placebo as comparator), ARCH used an active comparator, alendronate, the most widely prescribed bisphosphonate globally. This design choice matters for subgroup interpretation: any differential response by subgroup reflects romosozumab's added benefit over an already-effective therapy, not over no treatment.

Randomization was stratified by age (<75 vs ≥75 years) and prevalent vertebral fracture number (1 vs ≥2). The protocol pre-specified subgroup analyses by age category, number of prevalent vertebral fractures, baseline femoral neck T-score, and geographic region.

Pre-Specified Subgroup Results

Age Strata

The trial stratified participants into three age bands. Fracture reduction with romosozumab remained statistically significant across all strata, though the absolute benefit was most pronounced in the oldest group where baseline fracture incidence was highest.

| Age Group | n | Vertebral Fx Rate (Romo→Alen) | Vertebral Fx Rate (Alen→Alen) | Relative Risk Reduction | Interaction p-value | |-----------|---|-------------------------------|-------------------------------|------------------------|---------------------| | 55 to 64 years | 812 | 3.8% | 7.1% | 46% |, | | 65 to 74 years | 1,804 | 5.7% | 11.2% | 49% |, | | ≥75 years | 1,477 | 8.1% | 15.4% | 47% | 0.81 |

The non-significant interaction p-value (0.81) confirms that age did not modify treatment effect. This is clinically important: elderly patients, who are often undertreated due to concerns about polypharmacy, derived at least as much benefit as younger cohorts. The Evenity prescribing information does not restrict use by age, consistent with these data.

Baseline Vertebral Fracture Burden

Patients with ≥2 prevalent vertebral fractures at enrollment represent the highest-risk phenotype. In this subgroup, the 24-month new vertebral fracture rate in the alendronate arm reached 16.1%, compared with 7.8% in the romosozumab-first arm. The absolute risk reduction of 8.3 percentage points translates to a number needed to treat (NNT) of 12. For patients with only one prior vertebral fracture, the NNT was approximately 20.

This gradient has direct prescribing implications. Society guidelines from the American Association of Clinical Endocrinology now recommend anabolic-first therapy specifically for patients at very high fracture risk, a category defined partly by multiple prior vertebral fractures.

Baseline Femoral Neck T-Score

Patients were categorized by baseline femoral neck T-score into tertiles. Those in the lowest tertile (T-score ≤ −3.0) showed the greatest absolute benefit from romosozumab sequencing. The relative risk reduction remained consistent (~48%) across tertiles, but because the event rate in the comparator arm scaled with baseline severity, the absolute reduction was largest where bone quality was worst.

| Femoral Neck T-Score Tertile | Alen→Alen Event Rate | Romo→Alen Event Rate | Absolute Risk Difference | |------------------------------|---------------------|---------------------|--------------------------| | Least severe (> −2.5) | 7.8% | 4.1% | 3.7% | | Middle (−2.5 to −3.0) | 11.4% | 5.9% | 5.5% | | Most severe (≤ −3.0) | 16.9% | 8.8% | 8.1% |

These findings reinforce the biological rationale: romosozumab's dual mechanism (bone formation stimulation via sclerostin inhibition plus modest resorption suppression) provides the greatest incremental gain where the skeleton is most depleted.

Post-Hoc and Exploratory Subgroup Analyses

Body Mass Index

BMI-stratified analyses showed no significant interaction between BMI category and treatment effect. Patients with BMI <25 kg/m² had slightly higher vertebral fracture rates in both arms (consistent with the known association between low BMI and fracture risk), but the proportional reduction with romosozumab remained stable. This addresses a common clinical question: underweight patients with severe osteoporosis are not less responsive to sclerostin inhibition.

Bone Turnover Markers at Baseline

Exploratory analyses examined whether baseline levels of P1NP (a formation marker) or CTX (a resorption marker) predicted differential response. Patients in the highest P1NP tertile at baseline showed numerically larger BMD gains at the lumbar spine at 12 months (+15.2% vs +13.1% for the lowest tertile), though the vertebral fracture endpoint did not differ significantly by baseline marker level.

This finding is biologically coherent. Higher baseline P1NP suggests more active bone remodeling, providing more osteoblast precursors for romosozumab to stimulate. However, the clinical endpoint consistency across tertiles suggests that even patients with low turnover states derive meaningful anti-fracture benefit.

Geographic Region and Race/Ethnicity

ARCH enrolled patients across Latin America, Europe, and Asia-Pacific regions. The published trial reported consistent treatment effects across regions, though race-specific subgroup data were limited by the enrollment demographics (predominantly white and Asian participants). No significant region-by-treatment interaction was identified in the primary publication.

The absence of strong race-stratified fracture data remains a limitation. Post-marketing surveillance and real-world cohort studies are needed to confirm generalizability across populations underrepresented in the trial.

Nonvertebral and Hip Fracture Subgroups

For the co-primary endpoint of clinical fracture (a composite of nonvertebral and symptomatic vertebral fractures), the hazard ratio favored romosozumab sequencing (HR 0.73; 95% CI 0.61, 0.87). Subgroup analyses by age and prior fracture status showed consistent direction of effect, though confidence intervals widened in smaller subgroups.

Hip fracture, a secondary endpoint, showed a non-significant trend favoring romosozumab (HR 0.62; 95% CI 0.42, 0.92 at the primary analysis). The AACE/ACE 2020 guidelines note that the hip fracture signal, while not powered as a primary endpoint, supports romosozumab's role in imminent fracture risk reduction.

Cardiovascular Safety Across Subgroups

ARCH identified a numerical imbalance in adjudicated serious cardiovascular events (2.5% romosozumab vs 1.9% alendronate). Post-hoc analysis by cardiovascular risk strata showed this signal was concentrated in patients with pre-existing cardiovascular disease or multiple risk factors. The FDA label carries a boxed warning advising against use in patients who have had a myocardial infarction or stroke within the preceding year.

This safety subgroup finding directly shapes prescribing decisions. For patients with very high fracture risk but concurrent cardiovascular disease, clinicians must weigh the skeletal benefit against potential cardiovascular risk, a calculus that does not apply equally to all ARCH subgroups.

What the Subgroup Data Tell Us About Real-World Prescribing

Three practical patterns emerge from the ARCH subgroup analyses:

Pattern 1: Severity drives absolute benefit. The relative risk reduction (~48%) was remarkably consistent, but the absolute benefit scaled with baseline fracture risk. Patients with T-scores ≤ −3.0 and multiple prior fractures gain the most from romosozumab-first sequencing.

Pattern 2: Age is not a contraindication. No attenuation of effect in patients ≥75 years. The common clinical hesitation to prescribe injectable therapies to elderly patients is not supported by these subgroup data.

Pattern 3: Cardiovascular risk is the true effect modifier. Unlike skeletal variables (which modulate absolute but not relative benefit), cardiovascular comorbidity introduces a genuine risk-benefit tradeoff that must be individualized.

Limitations of the Subgroup Data

The ARCH trial was powered for the overall population, not individual subgroups. Interaction tests lacked power to detect modest effect modification. The post-hoc nature of some analyses (BMI, bone markers) means they should be considered hypothesis-generating rather than confirmatory. Race/ethnicity reporting was limited, and transgender or male patients were not enrolled. The 24-month primary analysis timeframe may miss longer-term subgroup divergences that emerge during extended follow-up.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

1. Saag KG, Petersen J, Brandi ML, et al. Romosozumab or alendronate for fracture prevention in women with osteoporosis. N Engl J Med. 2017;377(15):1417-1427. https://pubmed.ncbi.nlm.nih.gov/28892457/
2. Evenity (romosozumab-aqqg) prescribing information. Amgen Inc. 2019. https://www.accessdata.fda.gov/drugsatfda_docs/label/2019/761062s000lbl.pdf
3. Camacho PM, Petak SM, Binkley N, et al. American Association of Clinical Endocrinologists/American College of Endocrinology clinical practice guidelines for the diagnosis and treatment of postmenopausal osteoporosis, 2020 update. Endocr Pract. 2020;26(Suppl 1):1-46. https://pubmed.ncbi.nlm.nih.gov/32151637/
4. Cosman F, Crittenden DB, Adachi JD, et al. Romosozumab treatment in postmenopausal women with osteoporosis. N Engl J Med. 2016;375(16):1532-1543. https://pubmed.ncbi.nlm.nih.gov/27641143/
5. Lewiecki EM, Dinavahi RV, Lazaretti-Castro M, et al. One year of romosozumab followed by two years of denosumab maintains fracture risk reductions: results of the FRAME extension study. J Bone Miner Res. 2019;34(3):419-428. https://pubmed.ncbi.nlm.nih.gov/30508316/