In 2003, the first skeletal complication of aminobisphosphonate exposure was documented by our report of a boy with pamidronate (PMD)-induced osteopetrosis (OPT). More than 20 yr earlier, rickets and osteomalacia were known to follow excesses of etidronate, a first-generation bisphosphonate (BP). In 2004, osteonecrosis of the jaw, and in 2005, suppressed bone turnover, emerged as concerns for adults receiving amino-BPs.
We studied our patient in 2002 at 12 yr of age, 1.5 yr after cessation of intravenous doses of PMD. PMD had been administered by others for his unexplained episodic bone pains, claims of appendicular fractures during minor trauma, elevation of bone alkaline phosphatase (ALP) in serum, and “osteopenia” on DXA (spine Z-score, −1.0). However, our review of his extensive medical records indicated that OPT had developed in a previously normal skeleton. PMD had been infused in escalating doses and frequency probably because his bone pain seemed to respond, in keeping with early reports regarding osteogenesis imperfecta (OI), and his idiopathic hyperphosphatasemia improved, yet neither problem was corrected. While he received ∼2.8 g of PMD over 2.75 yr, bone modeling became suppressed, causing club-shaped deformities of his femurs and tibias, primary spongiosa (calcified cartilage) accumulated during endochondral bone formation, and spondylolysis appeared at L5—together signifying OPT. Two years after the BP was stopped, he fractured an osteosclerotic distal radius, suggesting a lingering toxic effect of the PMD.
Here, we report our re-evaluation of this patient 5 yr later and 6.5 yr after PMD exposure ended.
There is considerable controversy whether amino-BPs can cause oversuppression of bone remodeling in adults. In 2002, our patient showed that severe disruption of skeletal remodeling and modeling is possible from PMD in children. He acquired clinical, biochemical, radiographic, and histopathological features of OPT caused by cessation of osteoclast activity during exposure to this amino-BP. Accumulation of primary spongiosa and dysmorphic osteoclasts off bone surfaces documented and explained the resorption failure while he received a total dose of PMD approximately four times the amount emerging in the late 1990s as treatment for severe OI. Before the PMD infusions stopped at 10.5 yr of age, the dose had escalated to 100 mg once monthly. Accordingly, we recommended in 2003, as had been our practice, that bone modeling be monitored radiographically for children exposed to prolonged courses of BPs.
Amino-BPs have now been administered for many pediatric conditions. Nevertheless, optimum doses and durations for the skeletal disorders remain unclear, and withdrawal of treatment after several years followed by observation is becoming commonplace. The skeletal effects of BPs are known to linger, and PMD excretion in urine has been shown to continue up to 8 yr in children.
Now, our 2007 findings with this patient suggest that novel skeletal aberrations might develop in children when potent BP therapy ceases, especially if there has been significant impairment of bone modeling. As reviewed below, experience elsewhere may be emerging to support this concern.
The first publication regarding PMD for OI, the paradigm of pediatric osteoporosis, appeared in 1987. Radiographs of a girl given courses of PMD orally for 1 yr showed dense, parallel, metaphyseal bands.
In 1992, Van Persijn van Meerten et al. reported subtle metaphyseal undertubulation during amino-BP treatment for children with various disorders and “bone-within-bone” osteosclerosis after therapy. It was said that metaphyseal osteosclerosis resolved after treatment, with formation of new bone of normal density and original width. Of interest, the published radiograph of a knee of a boy with polyostotic fibrous dysplasia shows an expanded distal femoral metaphysis with a thin cortex during treatment, remarkably like what we encountered in our patient.
In 1996, Adami and Zamberlan cautioned that excessive BPs might impair bone modeling and remodeling and induce OPT-like changes in children, because this effect had been recognized a decade earlier in growing rats.
In 2002, Rauch et al. reported calcified cartilage accumulation and large osteoclasts in the iliac crests of children with OI treated with PMD, but no indication that this decreased remodeling caused clinical problems, although they stated that the possibility required close monitoring. Two years later, they examined the histology of PMD-induced, osteosclerotic bands in the iliac crest of a child with OI and concluded that this nascent bone containing primary spongiosa remodeled into bands of trabecular bone with less calcified cartilage.
Up to 2005, skeletal modeling defects within the long bones of OI patients receiving PMD were otherwise not encountered, or were sometimes illustrated without comment. Then, Ward et al. used radiographs in 2005 to evaluate metaphyseal modeling in children said to be receiving appropriate doses of PMD for localized bone diseases (e.g., osteonecrosis) and devised a “metaphyseal index” to quantitate physiologic “inwasting” in the distal femur. Their preliminary data indicated correct bone shaping. Nevertheless, it was recommended that metaphyseal modeling be evaluated in safety protocols. Also in 2005, Letocha et al. described improved vertebral geometry and spine BMD (by DXA and QCT) for a small number of OI children given PMD for 1 yr.
In 2006, Land et al. reported partial reconstitution of vertebral shaping during 2–4 yr of PMD therapy for moderate-severe pediatric OI. They concluded that the transverse metaphyseal bands from PMD persist ∼4 yr on average (range, 2–8 yr), consistent with normal bone remodeling. They noted that the femoral metaphyseal index was increased on average 26% after treatment. Whether there were clinical consequences of the bone expansion was judged unclear, with no evidence of adverse implications, and perhaps some benefit.
Also in 2006, Vallo et al. showed that PMD therapy for pediatric OI led to “size-adjusted” spine BMD increments that stabilized or sometimes waned after 2 yr of treatment. They questioned whether repeated cycles of PMD were necessary or would courses without intermediate therapy be of identical benefit.
With the exception of the 1992 paper by Van Persijn van Meerten et al., no mention is made in these publications of metaphyseal mineral content or shaping after BP withdrawal, because treatment was continuing or finishing.
Then, in 2006, Rauch et al. described spine BMD Z-scores in children and adolescents with moderate-severe OI 2 yr after 3–4 yr of cyclical PMD therapy. Bone resorption activity, judged by urine NTX levels, accelerated, but not to pretreatment values. The Z-scores decreased, although not nearly to pretreatment values, and fracture rates did not increase. Rates of change in BMD toward baseline were greatest in patients who continued to grow. Newly acquired bone appeared less dense radiographically compared with during treatment, both in vertebrae and in wrist metaphyses. Knees and post-treatment metaphyseal modeling changes were not commented on.
In 2007, Rauch et al. used CT to evaluate the distal radius of 23 OI children on average 1.9 yr after at least 3 yr (average, 5.8 yr) of PMD therapy. Z-scores for BMC decreased more in the metaphysis than diaphysis. The metaphyseal Z-score changes suggested that the osseous tissue added by growth after the last PMD infusion was probably as fragile as before treatment. Hence, BP therapy might be necessary until growth ceases, although they cautioned that the risk/benefit ratio was not yet clear. The junction of the old, dense bone and new, less dense bone might be the site of future fracture.
In 2007, Waterhouse et al. examined 17 children with various disorders on average 26 mo after a mean of 22 mo of treatment either with PMD or zolendronate. Knees were said not to show OPT modeling changes. They observed no adverse effects, including on lumbar spine radiographs.
Notably, however, Ward et al., in 2007, reported a girl with OI who showed a marked decrease to below baseline in volumetric BMD Z-scores in the lumbar spine and distal radius within 2 yr after stopping 4.5 yr of PMD therapy. They questioned whether BP treatment should be discontinued before completion of linear growth. Our inspection of their published radiographs showed metaphyseal widening.
Our patient still seems predisposed to fracture 6.5 yr after cessation of PMD. In 2005, Grissom et al. offered three mechanisms whereby fracturing could continue in children with OI despite PMD therapy: (1) BMD remains below fracture threshold, (2) bone fragility persists despite increased or even normal density, and (3) increased physical activity by the patient. Despite our patient's recurrent bone pain and enigmatic hyperphosphatasemia, we cannot say that a skeletal disease predating his PMD exposure is now causing some of his current findings. He does not have OI, and bone fragility seems to have occurred from BP-induced suppression of bone remodeling. We believe that toxic doses of PMD led to spondylolysis and forearm fractures in areas of dense bone. He may now be additionally prone to fractures because of metaphyseal expansion with osteopenia.
Skeletal modeling is increasingly understood to influence bone strength. Treatment of pediatric patients with BPs can disturb modeling because osteoclast-mediated resorption must continue for proper shaping of growing bones. Despite the improvement in remodeling shown by our patient's follow-up iliac crest histology, he acquired two modeling aberrations, documented best at his knees, that will now likely persist lifelong: metaphyseal expansion and regions of cortical thinning. Additionally, there are now metadiaphyseal notches in his distal femurs and proximal tibias, as well as aberrant shaping of his femoral necks. Why he has diminished cortical thickness in metaphyseal areas at his knees is unknown, but perhaps these reflect the increased bone circumference.
There is also evidence that abnormal bone remodeling persists in our patient. Some excess of calcified cartilage remains in his iliac crest, radiographically dense cortical bone is seen in diaphyses, and osteosclerotic periarticular bone is present in his knees. In fact, our patient sustained a pathologic fracture across an ulna at radiographically dense and widened diaphyseal bone, showing it to be weaker than the adjacent osteopenic, expanded, metaphyseal bone. This “chalkstick” fracture remains incompletely healed after 2 yr, perhaps reflecting some persisting PMD effect. Failure of skeletal resorption in OPT compromises bone quality because calcified cartilage accumulates and remodeling-mediated interconnection of osteons is impaired. Accordingly, we worry that he will continue to fracture through this dense, but poor-quality, diaphyseal bone. Recently, Weinstein et al., in a preliminary report, encountered the osteoclast dysmorphology seen in our patient's iliac crest in women who had received several years of alendronate orally. Return of our patient's osteoclast morphology to normal currently is somewhat reassuring. Clearly, our patient did not have a congenital OPT, but instead reversible osteoclast suppression. Improved remodeling might eventually restore normal BMD and quality at the osteosclerotic sites, but it seems unlikely the modeling defects will ever correct. Our patient warns that metaphyseal osteopenia may occur if antiresorptive therapy widens metaphyses and then stops as growth continues. In fact, this may be what Ward et al. (discussed above) described in 2007. Hence, CT or MRI should be especially important to investigate this concern in other BP-treated children and is being used to study some of our patients.
Before PMD exposure, the height, density, and endplate width of our patient's vertebrae did not appear diminished on radiographs, and his spine BMD Z-score was normal. When diagnosed with BP-induced OPT in 2002, he had developed end-plate thickening and an unhealed L5 spondylolysis, although bone scintigraphy showed no evidence of vertebral fractures. In 2007, after further linear growth without PMD exposure, there was a “bone-within-bone” (endobone) configuration in his vertebrae and throughout his skeleton. This finding occurs in genetic OPT and was reported after amino-BP withdrawal in children by van Persijn van Meerten et al. in 1992. However, our patient's vertebral endplates are now thin, trabecular bone peripheral and central to the osteosclerotic bands appears osteopenic, and the vertebral bodies have become somewhat rectangular. Spondylolysis (leading to spondylolisthesis) seems more prevalent in OPT and occurred during and several years after his PMD exposure. Notably, Aström et al. in 2007 reported an increased prevalence of spondylolysis and L5 spondylolisthesis after PMD treatment in infants with severe OI.
Our patient illustrates a pitfall in BMD interpretation when there is heterogeneous bone density. DXA assesses relatively large areas and can, therefore, be misleading. His spine, radius, and whole body BMD increased during PMD at an accelerated rate compared with age-matched controls and decreased to normal by age 17 yr despite areas of spinal and metaphyseal osteopenia and sclerotic “bone-within-bone” bands shown radiographically. “Neo-osseous osteoporosis” in widened metaphyses would likely go undetected by DXA that provides an “areal” (g/cm2) assessment of BMD.
Clinicians hope that biochemical markers of skeletal turnover will help prevent excessive suppression of bone remodeling when they use antiresorptive pharmaceuticals. For our patient, markers other than ALP gave no hint of skeletal disease 1.5 yr subsequent to PMD exposure. Despite his enigmatic hyperphosphatasemia, only serum osteocalcin was slightly increased in 2002. Urine NTX and free deoxypyridinoline (also hydroxyproline) were unremarkable then and again in 2007. In 2002, these markers emanated from elevated skeletal mass (total body BMD Z-score ∼ +2.3), perhaps explaining why low bone turnover would go undetectable. Of interest, however, serum BB-CK and TRACP activities in 2002 were consistent with OPT and are now unremarkable. This supports our impression that assaying both enzymes can help to monitor for ongoing BP toxicity.
In conclusion, follow-up of the first reported case of drug-induced OPT calls for continued study of long-term BP exposure in pediatric patients to know whether remodeling and modeling disturbances, including metaphyseal “neo-osseous osteoporosis,” will complicate cessation of therapy. This may occur if bone modeling has been suppressed and tubular bones are expanded. In these patients, both cortical bone thickness and trabecular bone porosity in metaphyses require evaluation. Currently, it is not clear if some level of BP treatment should continue for children until growth plates fuse.