Introduction
Denosumab [1] is a fully-humanized monoclonal
antibody against RANK-ligand, and a potent antiresorptive drug used for the
treatment of osteoporosis [2], effective in
increasing bone mineral density (BMD) and preventing osteoporotic fractures during,
at least, 10 years of continuous treatment [3]. Denosumab is administered at 6 months intervals. Soon after its
administration, bone resorption is markedly suppressed, as reflected by a decrease
in bone turnover markers (BTM) [4]. Bone
turnover suppression persists for 6 months and reverses rapidly thereafter [5]. If denosumab administration is not
reinstituted, a rebound acceleration of bone turnover occurs [6]. The pivotal FREEDOM trial demonstrated an
increase of bone turnover rate after a median follow-up of 2.4 months following the
would-be scheduled injection, associated with a decrease in BMD and increase in
vertebral fracture rate, similar to the fracture rate observed in the placebo group
[7]
[8].
Since 2015, attention was drawn to the possibility that denosumab discontinuation
could be linked to an increased risk for vertebral fractures, including multiple
vertebral fractures (MVF) [9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]. A mechanism involving rebound elevation of
bone turnover and an increased osteoclastogenesis was proposed [17], as well as an accelerated targeted repair
of accumulated microdamage [10]. A post-hoc
analysis of data from the FREEDOM trial and its extension revealed that denosumab
discontinuation was associated with a 6-fold increase in the rate of vertebral
fractures, approaching fracture rate in the placebo group, while the proportion of
multiple vertebral fractures was 1.6-fold higher in the denosumab discontinuation
group, more so in patients with prior vertebral fractures [18]. We obtained similar results, derived from
a real-world setting (a large health provider’s database), comparing
denosumab discontinuers and persistent users, and identified chronic kidney failure
and cerebrovascular disease as factors associated with fractures following denosumab
discontinuation [1]. It has been shown that
bone-resorption rate was attenuated in denosumab discontinuers treated with
bisphosphonates (BP) before denosumab [19],
but the evidence regarding such protection against rebound acceleration of bone
turnover and fractures is still controversial [9]
[15]
[16]
[20].
Bisphosphonates are the most recommended treatment modality post denosumab
discontinuation [21]. Some controversy exists
to whether a potent intravenous BP should be preferred or a more gradually acting
oral medication, with several reports supporting oral [22]
[23],
and others intravenous route [24]. Denosumab
treatment duration prior to discontinuation and individual fracture risk should be
taken into consideration as well [21].
Study objectives
In the present report, we provide a unique long-term observation on the
management and outcomes of patients with post-denosumab discontinuation MVF,
initially identified in a previous study from a large Israeli healthcare
provider’s database [1].
Materials and Methods
The study was performed using longitudinal data from the Maccabi Healthcare Services
(MHS) computerized database. MHS is the second largest healthcare provider and
insurer in Israel serving approximately 25% of the population with a
countrywide coverage. It maintains a central electronic database since 1998,
including medical records, laboratory test results, imaging, diagnoses, medical
procedures, referrals, prescriptions, and medication purchases. In order to be
reimbursed by the comprehensive national healthcare insurance plan, denosumab should
be mandatorily purchased in any of the pharm facilities associated with MHS and
recorded in its database. Data collection for our core study [1] ended by June 2018, and the data collection
for the current report ended by April 2020.
Patients were detected as previously described [1]. Denosumab initiators, with>2 consecutive medication purchases
starting January 2012, were identified. Treatment discontinuation was defined as a
refill gap of 3 months or more (>9 months after the last denosumab
dispensation). The date of an expected refill (6 months following the last refill)
was considered the discontinuation date. Patients with MHS membership of fewer than
12 months pre- or 15 months post denosumab initiation date were excluded.
Major osteoporotic fractures were categorized according to the MHS osteoporosis
registry, which includes vertebral, hip, distal forearm, and proximal arm fractures
[25]. All vertebral fractures, including
multiple vertebral fractures (MVF), were reviewed and adjudicated. Two or more
vertebral fractures were classified as MVF.
Fractures were defined as “post discontinuation” if they occurred
within 1 year from denosumab discontinuation.
Demographic and additional data was collected from electronic medical records and the
MHS oncology, cardiovascular or cerebrovascular registries, including osteoporosis
treatment before denosumab, data on traditional risk-factors for osteoporotic
fractures: body mass index (BMI), bone mineral density (BMD), alcoholic beverages
drinking habits, smoking, fracture history, secondary causes of osteoporosis
(rheumatoid arthritis, chronic or ever exposure to systemic glucocorticoids).
Medication possession ratio (MPR) of osteoporosis medications before denosumab
treatment was calculated from the electronic medical records and expressed as a
percentage of purchases per each year of treatment. Post-MVF osteoporosis
treatments, treatment duration, recurrent vertebral fractures, and fatalities, were
extracted. Recurrent vertebral fractures were considered “new” if
they occurred>12 months after the index MVF event.
Continuous variables are expressed as mean±standard deviation, or median and
interquartile range (IQR). Categorical data are expressed as counts and
percentages.
All procedures in this study followed the 1964 Helsinki declaration and its
amendments. The study was approved by the MHS Ethics Committee.
Results
The core study included 1500 denosumab discontinuers [1]. Twelve female patients with MVF were identified among the
discontinuers. The characteristics at denosumab initiation of the included patients
are presented in [Table 1]. MVF events
occurred a mean 3±1.2 denosumab injections and 134±76 days after
denosumab discontinuation. Of the 12 patients who sustained MVF, 3 fractured 2
vertebrae and 8 fractured 3 vertebrae. In one case, the precise number of fractured
vertebrae could not be determined.
Table 1 Characteristics at denosumab initiation of patients with
MVF.
|
Characteristics
|
Patients with MVF, n=12 among DD
|
|
Age, years, mean (SD)
|
71 (12)
|
|
Female,%
|
100
|
|
Prior fractures, n (%)
|
6 (50)
|
|
Prior vertebral fractures, n (%)
|
4 (33.3)
|
|
Prior hip fractures, n (%)
|
1 (8)
|
|
Prior non-hip-non-vertebral fractures, n (%)
|
3 (25)
|
|
Prior Osteoporosis treatment, n (%)
|
11 (91.6)
|
|
Denosumab injections prior to discontinuation, mean (SD)
|
3 (1.2)
|
|
Diabetes, n (%)
|
4 (33)
|
|
Rheumatoid arthritis, n (%)
|
2 (16.6)
|
|
Ever glucocorticoid use, n (%)
|
7 (58.3)
|
|
Current smoking, n (%)
|
2 (16.6)
|
|
Past smoking, n (%)
|
0
|
|
Femoral neck T-score, mean (SD)
|
–2.34 (0.8)
|
|
Lumbar spine T-score, mean (SD)
|
–2.42 (1.9)
|
|
BMI, kg/m2, mean (SD)
|
26.4 (3)
|
|
eGFR, ml/min/1.73 m2, mean (SD)
|
73 (27)
|
|
Cancer, n (%)
|
3 (25)
|
|
Cardiovascular disease, n (%)
|
4 (33)
|
|
Cerebrovascular disease, n (%)
|
3 (25)
|
MVF: Multiple vertebral fractures; DD: Denosumab discontinuers.
[Table 2] presents the prior fractures and
pre-denosumab medications details of the MVF patients. Osteoporotic fractures before
denosumab treatment occurred in 6/12 patients. All but one of the 12
described patients received other osteoporosis medications before denosumab. A total
of 10/12 patients were pre-exposed to bisphosphonates. One patient (P3)
received strontium ranelate, and one patient (P10) received raloxifene. One patient
(P7), completed a two years course of teriparatide 16 months before starting
denosumab. Mean MPR for oral BPs was 67.8±14.7%. Five patients with
prior BP exposure had less than one-year washout of BP before starting denosumab
treatment (grey-shaded rows). MPR in this subgroup was
69.8±15.7%.
Table 2 Prior osteoporotic fractures and pre-denosumab osteoporosis
treatment.
|
Patient (P)
|
Prevalent vertebral/hip fractures
|
Denosumab initiation year
|
Prior osteoporosis treatment, years, medication
|
Medication possession ratio, oral medications, average
(%)
|
|
P1
|
|
2013
|
1999–2007, oBP
|
59
|
|
P2
|
|
2015
|
2008–2015, oBP
|
96
|
|
P3
|
|
2013
|
2006–2012, oBP 2012–2013, Strontium
|
62 92
|
|
P4
|
Vertebral, hip
|
2016
|
2004–2005, oBP 2009–2012, oBP 2012, 2013, 2014,
IV Zoledronic acid
|
73
|
|
P5
|
Vertebral
|
2016
|
|
|
|
P6
|
|
2014
|
2008–2014, oBP
|
58
|
|
P7
|
Vertebral x 2
|
2015
|
2010–2011, oBP 2012–2014, Teriparatide
|
80
|
|
P8
|
|
2015
|
2010–2014, oBP
|
50
|
|
P9
|
|
2012
|
2000–2011, oBP
|
53
|
|
P10
|
|
2014
|
2011–2012, Raloxifene
|
25
|
|
P11
|
Vertebral
|
2014
|
2013, oBP
|
NA
|
|
P12
|
|
2015
|
2004–2014, oBP
|
72
|
oBP: Oral bisphosphonate; IV: Intravenous; NA: Not applicable. Grey-shaded
rows: less than 365 days between BP treatment and denosumab initiation.
[Figure 1] presents the post-MVF management
and clinical outcomes. The median follow-up duration after MVF was 36.5 (IQR
28.2–42.5) months. Six patients (P1–P4, P6, P9) resumed denosumab
treatment at some point following MVF, two of them did not persist with the
re-treatment (P3 and P6); one of them (P6) suffered an additional vertebral
fracture. One patient (P4) died soon (1 month) following the resumed treatment. Four
patients (P5, P7 and P9–10) received teriparatide as a first medication
following MVF, one of them (P9) was subsequently given an additional denosumab
injection and was then switched to zoledronic acid. One patient (P10) received oral
alendronate, one patient (P3) received oral risedronate alternating with denosumab,
and one (P8) received risedronate for 6 months, was subsequently switched to
zoledronic acid and then to teriparatide.
Fig. 1 Treatment allocations and clinical outcomes post MVF. Post MVF
Treatments: Medication (number of purchases); MVF- Multiple vertebral
fractures, DD- Denosumab discontinuation; Ris: Risedronate; Alnd:
Alendronate;*:Denosumab (1);**: MVF occurred two
months after treatment initiation; #: Recurrent vertebral fracture.
Two patients received no pharmacological treatment following MVF. One of those, died
14 months after the MVF event. Of note, two patients received teriparatide (P7) and
risedronate (P8) for two months following denosumab discontinuation, before MVF
occurred ([Fig. 1]).
Two patients underwent vertebroplasty/kyphoplasty (P5 and P10), four and
three months following the MVF event, respectively. In both cases the procedure was
performed while on teriparatide and no additional vertebral fractures were recorded
during 25 (P5) and 38 (P10) months of follow-up.
Discussion and Conclusions
We present patient-level data from a subgroup of patients with multiple vertebral
fractures (MVF), derived from a large real-world cohort of patients monitored during
one year for fractures following denosumab discontinuation. Our core study [1] demonstrated that any fracture, including
MVF, occurred more frequently in patients who discontinued denosumab treatment than
in persistent users, providing a real-world perspective on post denosumab
discontinuation fractures. Data presented in the current report concerns patients
with MVF and focused on existing osteoporosis risk factors, detailed prior
treatment, fractures timing, and long-term follow-up for post-MVF treatments as well
as clinical outcomes.
Remarkably, all our patients with MVF following denosumab discontinuation were women.
This corresponds with the vast majority of the previously published cases on
vertebral fractures following denosumab discontinuation [15]
[16]
[26]. We are aware of only two
reported male patients with vertebral fractures following denosumab discontinuation,
one of them with extremely low BMD, and the other with glucocorticoid-induced
osteoporosis [27]. Osteoporotic fractures,
although more prevalent in women, are reported in 1 in 3 men after 50, and the
prevalence of vertebral fractures in men>65 years old approaches half the
rate among women [28]. The reason for the
female dominance among the reported cases with vertebral fractures following
denosumab discontinuation is not clear. It could be due to a differentially higher
bone turnover rate in women during the first decade following menopause, which is
not present in men due to residual exposure to sex hormones [29], or due to structural and/or
biomechanical differences between male and female vertebrae [30]. The most plausible reason is the paucity
of denosumab prescriptions and anti-osteoporosis medications at large, in men in
general [31]. In our core study, among 1500
denosumab discontinuers, only 8 percent were male [1].
Prevalent VFs were documented in one-third of our MVF patients, which is not
different than what is expected in postmenopausal women, in whom prevalence of
vertebral fractures is reported to be between 20 and 29% in those over 75
[32]
[33]. Nevertheless, it should be noted that this historical prevalence
rates include asymptomatic vertebral deformities (about 2/3 of cases), while
most prevalent fractures in our core study were identified via coded diagnoses and
were therefore, in most part, clinically evident, suggesting that vertebral
fractures in our series could be more prevalent than in the general population. The
higher incidence of MVF in a subgroup with prevalent vertebral fractures among
denosumab discontinuers was previously emphasized [18], indicating increased baseline fragility of the vertebrae before
denosumab treatment, and implying that allocation to denosumab treatment may
comprise a patients’ selection bias.
It has been suggested that the rebound increase in bone resorption following
denosumab discontinuation was less pronounced in patients treated with
bisphosphonates before denosumab treatment [19], and MVF seemed to be less common [15], however, this is not necessarily generalizable. In a series of
high-risk osteoporotic patients with vertebral fractures following denosumab
discontinuation reported previously, all but one with MVF, long-term bisphosphonate
treatment before receiving denosumab was observed [16]. In the present study, a majority (10/12) of the MVF patients
were treated with oral and/or intravenous bisphosphonates for prolonged
periods before the initiation of denosumab treatment, with an MPR close to
70%. This suggests that bisphosphonate exposure before denosumab treatment
has only a limited protective effect, if any, against MVF following denosumab
discontinuation. The prevailing concept of over-suppression of bone turnover during
long-term oral bisphosphonate treatment has recently been challenged following
observations that a considerable proportion of the patients on long-term oral
bisphosphonates had a high level of bone resorption [34].
The number of fractured vertebrae per patient in our case-series is somewhat lower
than previously reported [15]
[16]
[35].
This could be due to a selection bias in previous case series of patients with high
severity at presentation, whereas our patients were systematically identified
following denosumab discontinuation [1].
Treatment during over 3 years of follow-up post MVF was very variable: patients were
prescribed denosumab, teriparatide, oral, and IV bisphosphonates in different
sequences. A third (2/6) of the patients who received denosumab following
MVF, discontinued treatment after a second treatment sequence.
Since a rebound increase in bone turnover rate is assumed to be the underlying
mechanism for MVF after denosumab discontinuation, a timely bone resorption
inhibition by reinstituting denosumab is perhaps the most appropriate treatment
modality [21], although, according to one
report, it may not completely eliminate the risk for recurrent vertebral fracture
[36]. In our core MHS study, most of the
cohort (1338/1500; 89%) discontinued denosumab treatment
administered before the end of 3 years [1].
This short-term exposure time was interpreted as implying inadvertent patient
dropout rather than planned physician-initiated discontinuation. If this is the
case, subsequent unintentional treatment discontinuations, or incompliance, are
equally likely and were indeed demonstrated in this sub-group analysis, indicating
that intensive patients’ education should be included as an integral part of
a comprehensive intervention for secondary prevention of osteoporotic fractures.
Suboptimal compliance is further suspected by incomplete treatment course with
teriparatide (P5, P7, P9, and P10) and lack of antiresorptive medication
administration following teriparatide (P5).
Only four patients (33%) of the denosumab discontinuers received
bisphosphonates, either oral or IV, after the MVF event, even though bisphosphonates
are the most commonly recommended treatment modality after denosumab discontinuation
[21]
[37]. A recently published systematic review and position statement by
European Calcified Tissue Society (ECTS) recommends that patients with shorter
duration of denosumab treatment (up to 2.5 years) and otherwise low fracture risk
should receive treatment with an oral bisphosphonate for 1 to 2 years. Patients who
have been treated with denosumab for a longer period or who are at persistently high
risk for fracture should receive zoledronate [21].
Four (a third) of our denosumab discontinuers received teriparatide as a first
medication after the MVF. As previously mentioned, rapid bone turnover suppression
is considered the cornerstone of treatment of the rebound increase in bone turnover
phenomena, while teriparatide exerts its action via stimulation of bone turnover
[38]. It has been shown that women who
received 2 years of denosumab and were switched to 2 years of teriparatide
experienced a transient BMD decline [39]. One
of our patients (P7) sustained MVF, after denosumab discontinuation and two months
into teriparatide treatment. It has been already suggested that teriparatide would
be undesirable following denosumab treatment [21]
[40].
Two denosumab discontinuers underwent vertebroplasty/kyphoplasty, procedures
previously shown, in several case reports, to bring about additional vertebral
fractures upon denosumab discontinuation and are currently not recommended for the
treatment of vertebral fractures in the scenario of denosumab discontinuation [35].
Two patients sustained recurrent vertebral fractures, one patient (P6) after
additional denosumab discontinuation and the other, P9, was treated with
teriparatide (though of the suboptimal duration), followed by denosumab and
zoledronic acid. Patient-related factors as well as vulnerability of the
already-fractured spine could have accounted for the scenario of the second patient
[41].
Over 36.5 months of follow-up, two patients in the MVF-DD group passed-away (one of
those was not treated following an MVF and another sustained recurrent vertebral
fracture), reflecting the extreme fragility of some of the patients and the
additional risk caused by recurrent fractures.
Our study has several limitations. Due to the retrospective nature of secondary data
collection from an HMO database, rather than from personal interaction with the
patients, we were unable to ascertain the rationale for the clinical decisions
regarding the management after denosumab discontinuation. Since reports on vertebral
fractures after denosumab discontinuation emerged after 2015, it is conceivable that
some laxity in the management of denosumab discontinuers, which in our patients
mainly occurred between 2015–2017, could have been affected by the lack of
clear practice guidelines regarding this condition at that time. Finally, a core
study design, specifically the definition of discontinuation as a 3-month (and not
less) refill gap and the follow-up for fractures post discontinuation as one year
(and not more) might have caused underestimation of fracture incidence and a
selection bias, since patients presenting before or after this time-frame may also
have fractures related to denosumab discontinuation, with distinct characteristics
.
The strengths of our report lay in the systematic methodology of data retrieval from
an established, high-quality, database, meticulous adjudication of the MVF events,
and unique long-term follow-up (>3 years) after MVF.
It should be noted, that the 2017 ECTS position statement [37], recommends that physicians, patients, and
regulatory authorities should be educated on the potential risk of discontinuing
denosumab treatment and on measures that should be taken to prevent unscheduled
treatment discontinuation. The 2019 Endocrine Society updated guidelines on
pharmacological management of osteoporosis in postmenopausal women, highlights the
need for alternative antiresorptive treatment upon denosumab discontinuation [42]. The 2020 updated version of the ECTS
position statement on fracture risk and management of discontinuation of denosumab
therapy [21], the authors further stress the
need for prompt intervention to reduce high bone turnover in patients who sustain
vertebral fracture following denosumab discontinuation.
In conclusion, we presented a case series of real-world patients with MVF following
denosumab discontinuation. The long-term follow-up reveals that management is
inconsistent, and recurrent fractures are not uncommon. It calls for special
attention to this high-risk group and implementation of management guidelines for
patients with MVF after denosumab discontinuation [21].
