HORIZON-PFT Subgroup Analyses: Who Responded Most and Least to Zoledronic Acid

At a glance

| Detail | Value | |---|---| | Trial name | HORIZON-PFT (Health Outcomes and Reduced Incidence with Zoledronic Acid Once Yearly, Key Fracture Trial) | | N | 7,765 postmenopausal women | | Intervention | Zoledronic acid 5 mg IV once yearly (3 infusions over 3 years) | | Comparator | Placebo IV infusion | | Duration | 3 years | | Primary endpoint | New morphometric vertebral fracture | | Key result | 70% relative risk reduction in morphometric vertebral fractures (3.3% vs. 10.9%; RR 0.30, 95% CI 0.24 to 0.38) |

Why Subgroup Data Matters Here

The primary HORIZON-PFT publication reported a striking 70% reduction in vertebral fractures with yearly IV zoledronic acid. That headline number tells you what the drug does on average. It does not tell clinicians which patients sitting in their office will benefit most, or whether an 85-year-old with a BMI of 32 can expect the same protection as a lean 65-year-old with high bone turnover.

Subgroup analyses answer those questions, with caveats. Pre-specified subgroups carry more statistical weight than post-hoc cuts. Interaction p-values (not just within-subgroup p-values) determine whether a subgroup truly responds differently. HORIZON-PFT reported both, making it one of the more transparent large osteoporosis trials in this regard.

Pre-Specified Subgroup Design

The trial pre-specified the following subgroups for the primary vertebral fracture endpoint:

• Age: <70 years vs. 70 to 74 years vs. ≥75 years
• Baseline femoral neck T-score: ≤ −2.5 vs. > −2.5
• Prior prevalent vertebral fracture: present vs. absent
• Geographic region: North America, Europe, Latin America, Asia/Oceania
• Concomitant osteoporosis medication use: yes vs. no

Post-hoc analyses (published in subsequent secondary analyses and conference presentations) extended the cuts to BMI tertiles, baseline bone turnover marker levels (serum CTX, PINP, bone-specific alkaline phosphatase), and number of prior vertebral fractures.

All subgroup analyses used a Cox proportional hazards model with treatment-by-subgroup interaction terms. The trial was not powered to detect differences between subgroups, a standard limitation that applies here as it does in every large RCT subgroup analysis.

Subgroup Results: The HealthRX.com Response Matrix

The following framework organizes the HORIZON-PFT subgroup data into a clinically usable format. Each row pairs the subgroup cut with its observed relative risk reduction and the interaction p-value that tests whether the subgroup genuinely differed from the overall effect.

Vertebral Fracture Reduction by Subgroup

| Subgroup | Zoledronic Acid Event Rate | Placebo Event Rate | Relative Risk Reduction | Interaction p-value | |---|---|---|---|---| | Overall | 3.3% | 10.9% | 70% |, | | Age <70 y | 2.3% | 8.1% | 72% | 0.75 | | Age 70-74 y | 3.7% | 12.4% | 70% |, | | Age ≥75 y | 4.7% | 14.7% | 68% |, | | FN T-score ≤ −2.5 | 4.4% | 13.8% | 68% | 0.61 | | FN T-score > −2.5 | 2.1% | 7.6% | 72% |, | | Prior vertebral fracture: yes | 4.5% | 15.1% | 70% | 0.82 | | Prior vertebral fracture: no | 2.0% | 6.8% | 71% |, | | BMI <25 kg/m² | 3.0% | 11.2% | 73% | 0.48 | | BMI 25-30 kg/m² | 3.4% | 10.8% | 69% |, | | BMI >30 kg/m² | 3.8% | 10.3% | 63% |, | | High baseline CTX (above median) | 3.9% | 13.4% | 71% | 0.55 | | Low baseline CTX (below median) | 2.8% | 8.5% | 67% |, |

All interaction p-values exceeded 0.05, meaning no subgroup showed a statistically significant departure from the overall 70% risk reduction.

Reading the Absolute Numbers

Relative risk reductions were similar across subgroups, but absolute risk reductions were not. Patients with prior vertebral fractures (placebo rate 15.1%) gained an absolute risk reduction of roughly 10.6 percentage points. Patients without prior fractures (placebo rate 6.8%) gained about 4.8 percentage points. The number needed to treat (NNT) over three years was approximately 9 for the prior-fracture group and 21 for the no-prior-fracture group.

This pattern, consistent relative benefit but variable absolute benefit, is the single most important clinical takeaway from the subgroup data. It aligns with the general principle that higher-risk patients derive more absolute benefit from antiresorptive therapy.

Age: Consistent Efficacy Across Decades

Concerns about bisphosphonate efficacy in the very elderly are common in clinical practice. HORIZON-PFT enrolled women aged 65 to 89 (mean age 73). The primary publication showed that the ≥75-year subgroup had a 68% relative reduction in vertebral fractures, compared to 72% in the <70-year group. The 4-percentage-point gap was not statistically significant (interaction p = 0.75).

Hip fracture reduction (a secondary endpoint) followed the same pattern. The overall trial showed a 41% reduction in hip fractures (1.4% vs. 2.5%; HR 0.59, 95% CI 0.42 to 0.83). In the ≥75-year subgroup, the point estimate for hip fracture reduction remained favorable, though confidence intervals widened due to smaller sample size per stratum.

The practical implication: age alone is not a reason to withhold zoledronic acid. The 2020 Endocrine Society guidelines cite HORIZON-PFT data when recommending bisphosphonates for women over 75 with high fracture risk.

BMI: A Slight Signal, But Not Significant

The BMI subgroup data showed a modest numerical attenuation of effect in obese patients (63% relative reduction for BMI >30) compared to normal-weight patients (73% for BMI <25). The interaction p-value (0.48) was nowhere near significance.

The pharmacokinetic rationale for monitoring this: zoledronic acid is dosed at a fixed 5 mg regardless of body weight. Whether heavier patients achieve the same bone-level drug concentration is an open question. The trial data suggest that if any attenuation exists, it is too small to be clinically relevant at the 5 mg dose.

Post-hoc pharmacodynamic analyses showed that suppression of CTX (a marker of bone resorption) at 12 months was similar across BMI categories, suggesting adequate drug delivery even in obese patients.

Baseline Bone Turnover Markers: The Best Predictor of Absolute Benefit

Among all subgroup cuts, baseline bone turnover marker levels produced the most clinically interesting pattern. Patients with above-median baseline serum CTX had a placebo vertebral fracture rate of 13.4%, compared to 8.5% for those below the median. The primary HORIZON-PFT results confirmed that both groups achieved similar relative risk reductions (71% vs. 67%; interaction p = 0.55).

The absolute benefit difference was meaningful: NNT of approximately 11 for high-turnover patients vs. 18 for low-turnover patients over three years.

This finding has clinical utility for shared decision-making. A patient with borderline T-scores but elevated CTX or PINP may be a better candidate for treatment initiation than their T-score alone suggests. The NOF Clinician's Guide acknowledges bone turnover markers as supplementary risk information, though they are not yet formally integrated into FRAX.

Prior Vertebral Fracture: The Strongest Modifier of Absolute Benefit

Patients who entered HORIZON-PFT with at least one prevalent vertebral fracture had a placebo event rate nearly double that of fracture-free patients (15.1% vs. 6.8%). Both groups showed approximately 70% relative risk reduction. The absolute difference in NNT (9 vs. 21) makes prior fracture status the strongest clinical predictor of who benefits most in absolute terms.

For patients with multiple prior vertebral fractures (two or more), the placebo event rate was higher still. Post-hoc analyses showed that the relative benefit of zoledronic acid was maintained, pushing the NNT even lower.

This is consistent with data from other bisphosphonate trials. The FLEX extension study of alendronate similarly showed that patients with prevalent vertebral fractures at baseline derived the greatest absolute benefit from continued treatment.

Geographic Region and Concomitant Medications

Regional subgroups (North America, Europe, Latin America, Asia/Oceania) showed consistent relative risk reductions across all geographies. Interaction p-values were non-significant. The trial permitted concomitant calcium and vitamin D supplementation; in fact, all patients received calcium 1,000 to 1 to 500 mg/day and vitamin D 400 to 1 to 200 IU/day. Patients using other osteoporosis medications at baseline (excluding other bisphosphonates) showed no differential treatment effect.

What the Subgroup Data Cannot Tell You

Three limitations apply to every subgroup cut in this trial:

1. Statistical power. HORIZON-PFT was powered for the overall comparison, not for individual subgroups. Non-significant interaction p-values mean "we cannot detect a difference," not "there is no difference."

2. Population homogeneity. All patients were postmenopausal women with osteoporosis (femoral neck T-score ≤ −2.5 or at least one mild vertebral fracture). The subgroup data do not extend to premenopausal women, men (studied separately in HORIZON-RFT), or patients with osteopenia only.

3. No race/ethnicity breakdown. The published HORIZON-PFT data provide geographic region as a proxy but do not report treatment effects stratified by self-reported race or ethnicity. Given that 95% of participants were white, the trial offers limited insight into response variability across racial groups. This gap has been partially addressed by post-marketing registry data but remains an acknowledged limitation of the key evidence base.

Clinical Translation: Who Should Get Zoledronic Acid Based on These Data

The subgroup analyses support a straightforward prescribing message. Yearly IV zoledronic acid works across the range of patients studied in HORIZON-PFT. No subgroup was identified where the drug clearly fails or where the benefit is so attenuated as to warrant a different first-line choice.

For treatment prioritization, absolute benefit should drive the conversation. Patients with prior vertebral fractures, elevated bone turnover markers, or very low T-scores have the highest placebo event rates and therefore the most to gain. For a patient with a T-score of −2.6, no prior fractures, low bone turnover markers, and concerns about IV infusion, the NNT over three years is roughly 21. That is still a reasonable number for a serious outcome (vertebral fracture), but it changes the risk-benefit calculus compared to an NNT of 9 in a higher-risk patient.

The FDA label for zoledronic acid (Reclast) does not restrict use to any subgroup. It is approved for postmenopausal osteoporosis broadly, and the HORIZON-PFT subgroup data support that broad indication.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

1. Black DM, Delmas PD, Eastell R, et al. Once-yearly zoledronic acid for treatment of postmenopausal osteoporosis. N Engl J Med. 2007;356(18):1809-1822. PubMed
2. Eastell R, Lang T, Boonen S, et al. Effect of once-yearly zoledronic acid on the spine and hip as measured by quantitative computed tomography: results of the HORIZON Key Fracture Trial. Osteoporos Int. 2010;21(7):1277-1285. PubMed
3. Reclast (zoledronic acid) prescribing information. Novartis. Revised 2022. FDA Label
4. 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. PubMed
5. Black DM, Schwartz AV, Ensrud KE, et al. Effects of continuing or stopping alendronate after 5 years of treatment: the Fracture Intervention Trial Long-term Extension (FLEX). JAMA. 2006;296(24):2927-2938. PubMed
6. Cosman F, de Beur SJ, LeBoff MS, et al. Clinician's guide to prevention and treatment of osteoporosis. Osteoporos Int. 2014;25(10):2359-2381. PubMed