Clinical characteristics of pregnancy and lactation associated osteoporosis: An online survey study

Abstract

Summary

Pregnancy and lactation associated osteoporosis is a rare and often severe osteoporosis presentation. Little information is available about etiology, clinical characteristics, risk factors and predictors of severity. Using an anonymized questionnaire, we defined clinical characteristics and potential risk factors for disease severity in PLO including primiparity, heparin exposure and celiac disease.

Purpose

Pregnancy and lactation associated osteoporosis (PLO) is a rare form of early-onset osteoporosis in which young women present with fractures, usually multiple vertebral fractures, during late pregnancy or lactation. Little information is available about etiology, clinical characteristics, risk factors and predictors of disease severity.

Methods

PLO patients were recruited to complete an anonymized online questionnaire. Disease severity was defined as total number of fractures during or after the first pregnancy associated with a fracture(s). Analyses related disease severity to potential predictors including diseases/conditions or medication exposures.

Results

177 completed surveys were received between 5/29/2018 and 1/12/2022. Average age at initial PLO fracture event was 32 ± 5 years. The majority were primiparous with singleton pregnancy and 79% fractured during lactation. Subjects reported 4.7 ± 2.7 total PLO fractures, with 48% reporting ≥ 5 fractures. Vertebral fractures, reported by 164/177 responders (93%), were the most common fracture type.

Conditions and medications most commonly reported included vitamin D deficiency, amenorrhea unrelated to pregnancy, nephrolithiasis, celiac disease (CD), oral steroid use, heparin products during pregnancy and progestin only contraceptive after pregnancy. CD and heparins exposure during pregnancy were significantly related to disease severity.

Conclusion

This is the largest study characterizing clinical features of PLO to date. The large number of participants and broad range of clinical and fracture characteristics queried has yielded novel information on the characteristics of PLO and potential risk factors for its severity, including primiparity, exposure to heparin and CD. These findings provide important preliminary data that can help target future mechanistic investigations.

Introduction

Pregnancy and lactation associated osteoporosis (PLO) is a rare, often severe, form of early-onset osteoporosis in which young women develop low trauma fractures during late pregnancy or lactation. Information on clinical characteristics of this rare condition is largely derived from case reports, as well as a few cohort studies. The most common presenting symptoms include severe back pain and height loss with a majority of women presenting with vertebral fractures [1–6]. Prior smaller cohort studies and case reports show that most cases occur in primigravid [1, 3, 6] women who are otherwise healthy and have no known predisposing conditions [1, 3, 6]. Pre-pregnancy bone mineral density (BMD) is usually unknown in these patients, as there would have been no indication to measure it. However, BMD measured by dual energy x-ray absorptiometry (DXA) at time of presentation is generally very low with Z scores often below −3.0 [3, 5, 6]. Prior cohort studies [1, 4, 7] have excluded those with known predisposing conditions, limiting ability to fully investigate potential mechanisms of this condition.

Substantial adaptations of the maternal skeleton and bone metabolism occur during normal pregnancy and lactation, which are states of high calcium demand due to fetal and infant calcium needs. Skeletal demineralization and maternal bone loss meet some of these needs during normal reproduction. Thus, normal pregnancy and lactation are associated with rapid asymptomatic decreases in spine and hip BMD—3–5% losses over pregnancy and an additional 3–10% loss over 6 months of lactation—followed by recovery [2, 8–10].

Even though temporary reductions in BMD are to be expected during this period, fractures and persistently low BMD are rare and distinguish women with PLO. Since most women with PLO are evaluated and diagnosed with osteoporosis in the context of their presentation postpartum, their skeletal status before pregnancy is unknown. Therefore, it remains possible that women with PLO have pre-existing skeletal fragility that is unmasked with the added stress of normal metabolic bone changes during pregnancy. It is also possible that women with PLO develop bone fragility due to unusual skeletal responses to the metabolic stress of pregnancy. Further research is needed to better understand potential mechanisms and etiologies of this condition.

In 2017, we initiated a cross-sectional study funded by the FDA Orphan Products Natural History Studies Grants Program to investigate clinical, hormonal, metabolic, bone structural and genetic characteristics of PLO. Herein, we present data addressing the first aim of the study: to define self-reported clinical characteristics of PLO utilizing an anonymized online questionnaire completed by affected patients. We aimed to learn more about the clinical characteristics of PLO, as well as clinical factors and predictors that may contribute to disease severity.

Methods

Between 5/2018 and 1/2022, subjects with PLO were recruited by advertisement, self- and physician-referral, and postings on online PLO support group sites. Any woman who self-identified as having PLO was eligible for the study. The questionnaire provided statements defining PLO eligibility: Subjects were invited to complete the survey if they had a fracture during pregnancy, within 12 months of delivery, while breastfeeding or within 6 months of stopping breastfeeding. Fracture type, timing and associated level of trauma were assessed. Eligible fractures included those of any bone (excluding face, skull, digits) associated with no trauma or trauma equivalent to a fall from standing height or less. Women who did not report fractures defined by timing, as described above, were excluded. Women could participate regardless of current menopausal status, prior osteoporosis treatment or history of known (secondary) causes of bone fragility. Individual survey identifiers prevented inclusion of duplicate subject data. Surveys with incomplete individual survey identifier data were excluded. The survey was distributed via web-link in a confidential manner. Subjects were offered the option to provide name and contact information. However, survey data remained confidential and separated from name and contact information.

The online survey was developed using Qualtrics Survey Software (Qualtrics XM, Qualtrics Inc, Provo, UT), licensed to Columbia University. The confidential online survey was designed to take approximately 30 min to complete. Question types included: multiple choice, numerical response, and text responses.

The survey collected data on demographic and reproductive characteristics of the subjects, characteristics of the PLO event(s) including fracture number, type and location, timing in relation to pregnancy and breastfeeding, and associated level of trauma for each fracture(s), BMD data and treatments received. Subjects were able to document up to 7 fracture events and could add text responses about fracture events in excess of 7. Trauma level was defined for each fracture event based on a multiple-choice question regarding the type of trauma or injury that was associated with the fracture. Response options for this question included: (1) The fracture event occurred with HIGH TRAUMA (for example motor vehicle accident, fall down flight of stairs), (2) The fracture event occurred with MINIMAL TRAUMA (for example: fall from a standing height, slip on the ice), (3) The fracture event occurred while performing normal daily activities (walking, stretching, lifting), (4) The fracture event occurred in the context of running or exercising – this may have been called a stress fracture, (5) There was NO trauma, injury or specific activity associated with the fracture, (6) Fractures(s) were seen on x-ray or other imaging, but it is not clear how or when they happened.

The survey also queried participants regarding multiple clinical conditions and medication exposures hypothesized to influence bone health or to be potential secondary causes of osteoporosis or bone fragility, including: endocrinopathies (Cushing syndrome, thyroid over-function, primary hyperparathyroidism), anorexia nervosa, exercise-induced amenorrhea, certain gastrointestinal conditions (celiac disease, ulcerative colitis, Crohn disease), liver diseases, cystic fibrosis, Gaucher disease, inflammatory conditions (e.g. lupus, rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis, other inflammatory arthritis), diabetes (gestational, type 1 or 2), malignancy, renal disease, renal stones, alcohol use and smoking. Medication exposure was assessed for various glucocorticoids (including duration of use), heparin, low molecular weight heparin (LMWH), depo medroxyprogesterone acetate (DMPA; Depo-Provera), progestin-only oral contraceptives after pregnancy, anti-epileptics, and anti-cancer medication. Data were also collected on exposure to low dose estrogen oral contraceptive pills (OCPs), selective serotonin reuptake inhibitors (SSRIs) and prescription stimulant usage (methylphenidate hydrochloride, amphetamine).

The Columbia University Institutional Review Board (IRB) reviewed and approved this study. A waiver of written documentation of consent was approved by the IRB. An IRB-approved information sheet which captured the elements of the consent but did not require participant signature preceded the survey.

Statistical analysis

Analyses were performed using SPSS (IBM Corp., Version 27, Armonk, NY). Characteristics of the included subjects are presented as mean ± SD or number of subjects and percent of responders to a given question.

Disease severity was defined as the total number of PLO fractures during or after the first pregnancy associated with a PLO event. Based on pre-specified hypotheses, we tested the following potential predictors of disease severity:

BMI, age at menarche, family history of osteoporosis, childhood fractures, age at first PLO fracture event, pregnancy characteristics (including parity and recommendation for activity restriction), PLO fracture site (e.g. vertebral, hip), initial PLO fracture timing (during pregnancy vs during lactation, and number of months postpartum).

Analyses were also conducted to relate disease severity to any disease/condition or medication exposure reported by 5 or more subjects. Since not all variables were normally distributed (Kolmogorov–Smirnov test), Spearman correlation analyses were used to describe relationships between disease severity and continuous variables. Relationships between disease severity and categorical variables were assessed using Mann–Whitney U test.

For conditions/medication exposures significantly related to disease severity, between groups comparisons were conducted to define demographic, reproductive and fracture characteristic differences between those with and without the reported potential risk factor. Between groups comparisons utilized Student’s t tests for continuous variables and chi square test or Fisher’s exact test for categorical variables.

All data are expressed as mean ± standard deviation (SD). Median and interquartile range (IQR) is provided for nonparametric analyses. Results were considered significant with p < 0.05.

Results

177 completed surveys were received between 5/2018 and 1/2022. Participant characteristics are shown in Table 1. The average age at survey completion was 36.4 ± 7.7 years. A majority of participants identified as White (91.4%) and Non-Hispanic (93%). The average BMI was 21.9 kg/m² based on self-reported height and weight. Most were born in North America (76), followed by Europe (71), Australia (18), Asia (4), Africa (2) and South America (1).

Table 1.

Clinical Characteristics and PLO Fracture Characteristics of the 177 Survey Participants

Clinical Characteristics	Mean ± SD or number reporting (percent)*
Age at time of survey participation (years)	36.4 ± 7.7
Race/Ethnicity	91% White/93% Non-Hispanic
Height (cm) at time of survey	166.8 ± 22.8
Weight (kg) at time of survey	60.2 ± 13.6
BMI (kg/m2)	22.3 ± 4.6
Age at menarche (years)	13.0 ± 1.6
Age of first pregnancy (years)	29.7 ± 5.6
Family history of osteoporosis	91/175 responders (52%)
History of childhood fracture(s)	74 (42%)
Number of fracture events in adulthood	1.2 ± 0.6
Total number of adult fractures (including traumatic fractures)	5.1 ± 3.0
Low trauma fracture prior to PLO event	13 (7%)
Known history of bone fragility/osteoporosis before PLO event	8/175 responders (5%)
BMD measurement prior to the event	11 (6%)
Characteristics of PLO Fractures
Age at first PLO fracture(s)	32.4 ± 5.0
Pregnancy characteristics
Primiparous	118/176 (67%)
Pregnancy with multiples (eg twin pregnancy)	6/176 (3%)
Activity restriction/bedrest recommended during pregnancy	32/176 (18%)
Infant birthweight (grams)	3428 ± 478
Total # of PLO Fractures	4.7 ± 2.7
Number of PLO Fractures
Less than 5 fractures	93 (52%)
5–10 fractures	80 (46%)
> 10 fractures	4 (2%)
PLO Fracture Site
Vertebral only	153 (86%)
Vertebral and other(s)	8 (5%)
Hip only	5 (3%)
Hip and other(s)	3 (2%)
Vertebral and hip	3 (2%)
Fractures other than vertebral and hip	5 (3%)
Trauma/injury associated with PLO Event
High trauma	6
Minimal trauma	21
While performing normal daily activities	67
While exercising/stress fracture	3
No trauma or injury	37
Unknown timing (fractures found on imaging)	40
PLO Fracture Timing
Fracture during pregnancy	26 (15%)
Among those reporting fracture during pregnancy:
Months gestation at time of event	7.0 ± 1.9
Fracture during lactation	140 (79%)
Among those reporting fracture during lactation:
Months postpartum at time of event	1.9 ± 1.9
Reported feedings/day at time of event
6 or more feedings	129
3–5 feedings	14
1–2 feedings	1
Delay in diagnosis reported	160 (90%)
# of weeks delay (weeks)	12.4 ± 12.3
Height loss
Bone Mineral Density (BMD) Assessment After the PLO Event
Reported BMD T scores
Lumbar Spine	−3.4 ± 1.1 (n = 150)
Total Hip	−2.1 ± 1.0 (n = 118)
Femoral Neck	−2.2 ± 1.0 (n = 99)
Age at time of BMD measurement (years)	32.8 ± 4.9
Timing of BMD measurement (months after delivery)	5.6 ± 5.4
BMD assessment while breastfeeding	163/169 (96%)

^*Number of responders (denominator for categorical variables) is 177 unless otherwise indicated

Family history of osteoporosis and history of childhood fractures were common, reported in 51% and 42%, respectively, and 7% reported having had low trauma fractures before the PLO fracture. Only 5% had a known history of bone fragility or osteoporosis before the PLO event, and only 6% had a BMD measurement prior to the event. Among the 6 subjects with BMD measurements before the PLO fracture(s), reported BMD T score at the spine averaged −2.6 ± 1.0.

Fracture characteristics

Fracture characteristics are shown in Table 1. Average age at the time of the initial PLO fracture was 32 ± 5 years. Most affected women (67%) were primiparous with singleton pregnancy (97%). Activity restriction during pregnancy was reported by 18%.

The number of fractures reported ranged from one to 12; most women reported multiple fractures (Fig. 1). On average, participants reported 4.7 ± 2.7 fractures with 48% reporting 5 or more fractures. Vertebral fractures were most common, reported by 164/177 (93%) women, with 153/177 (86%) reporting only vertebral fractures. Hip fractures were reported by 11/177 (6%) women, of whom 2% reported only hip and 2% reported hip and other types of fractures. Both vertebral and hip fractures were reported by 4/177 subjects (2%). Non-vertebral and non-hip fractures were reported by 5/177 (3%). Fractures were associated with 5 levels related to trauma: Fractures occurred in 21 women with minimal trauma, defined as fall from standing height. In 38 women, fractures were not associated with any trauma, injury or specific activity, while 67 fractured while performing normal daily activities, such as walking, lifting, and stretching. In 41 women, fractures were found with imaging. Three women fractured while running or exercising. Of the 6 women who categorized their fractures as high trauma, 3 fractured during manipulation by chiropractor/osteopath and 3 fractured during active labor. These fractures were included as PLO fractures.

Fig. 1.

Number of survey responders reporting each quantity (number) of fractures associated with a PLO event. Number of fractures reported with a single PLO event ranged from one to twelve

The majority of participants had fractures during lactation (79%), with only 15% occurring during pregnancy. Fractures during lactation occurred an average of 1.9 ± 1.9 months after delivery. The majority of women were breastfeeding ≥ 6 times/day (n = 129 reporting) when the fracture(s) occurred. Fractures during pregnancy occurred an average of 7.0 ± 1.9 months gestational age.

Notably, 91% of participants (160/175) experienced a long delay before being diagnosed with PLO fractures. On average, women reported a delay of 12.4 ± 12.3 weeks between the onset of pain and PLO fracture diagnosis. Height loss was reported in 61% of subjects.

Nine women sustained additional fractures coinciding with subsequent pregnancies. Of those nine women, eight reported between 1–10 vertebral fractures and one reported a hip fracture. Seven women reported subsequent fractures of the ribs, pelvis, foot, vertebrae, hips and forearm, which were unrelated to pregnancy. While we surveyed participants about subsequent fractures during pregnancy, we do not have data regarding subsequent pregnancies of all participants to assess rate of subsequent fractures. Of the 9 women who had subsequent PLO fractures, 8 had fractures in their first and second pregnancies while one had fractures in second and third pregnancy. An additional 5 women had subsequent fractures that were not related to pregnancy.

Participants reporting hip fracture

Eleven women (6%) reported a hip fracture, the majority during pregnancy (91%). Of this group, 100% were primiparous and 55% reported activity restriction during pregnancy (before or after the fracture event). BMI in the overweight range was reported by 36%; mean BMI (25.6 ± 6.9 kg/m²) was slightly higher than those without hip fracture (22.1 ± 4.4 kg/m²), but the difference was not significant (p = 0.1).

Fracture type in relation to fracture timing

Figure 2 shows the proportion of hip and vertebral fractures reported during pregnancy vs during lactation. Of note, while the majority of hip fractures occurred during pregnancy, the majority of vertebral fractures occurred during lactation.

Fig. 2.

Fracture timing in relation to pregnancy versus lactation. Vertebral fractures were more commonly reported during lactation and hip fractures were more commonly reported during pregnancy

BMD

The majority of participants (95.5%) had BMD measured after they fractured, predominantly during lactation (96%). BMD by DXA was measured an average of 5.6 ± 5.4 months after delivery. In those who reported BMD results, T scores (Table 1) averaged −3.4 ± 1.1 at the spine (n = 150), −2.1 ± 1.0 at the total hip (n = 118) and −2.2 ± 1.0 at the femoral neck (n = 99). Based on (postmenopausal) T score criteria, at the lumbar spine, 0.7% had normal BMD, 16% were in the osteopenic range and 83% were in the osteoporotic range. At the total hip, 8% had normal BMD, 58% were in the osteopenic range and 34% were in the osteoporotic range. At the femoral neck, 6% had normal BMD, 46% were in the osteopenic range and 44% were in the osteoporotic range. Of the patients with hip fractures, 33% were in the osteoporotic range at the lumbar spine (2/6 subjects), 67% at the total hip (2/3 subjects) and 25% at the femoral neck (1/4 subjects).

Osteoporosis medications

A wide array of osteoporosis treatments were reported. Teriparatide (29/171) and alendronate (14/166) were the most commonly reported treatments.

Medical conditions

Medical conditions reported are shown in Table 2. Conditions reported by 5 or more participants included vitamin D deficiency (77 patients, 43%), amenorrhea in excess of 3 months (41 patients, 23%), renal stones (13 patients, 7%), celiac disease (11 patients, 6%), gestational diabetes (10 patients, 6%), hyperthyroidism (6 patients, 3%), any cancer diagnosis (6 patients, 3%), liver disease (5 patients, 3%) and exercise-induced amenorrhea (5 patients, 3%). Cancers reported included breast (2), ovarian, thyroid and skin (2). Among the 13 participants with nephrolithiasis, 5 experienced symptoms during pregnancy and 8 had a family history of renal stones.

Table 2.

Medical History and Medication Exposures Potentially Related to Osteoporosis

	Number reporting/Number of responders (percent)*
Reported History of Diagnosis
Endocrine Conditions
Cushing syndrome	0
Hyperthyroidism (thyroid over-function)	6/176 (3%)
Primary hyperparathyroidism	2 (1%)
Vitamin D deficiency (past or current)	77/176 (43%)
Anorexia Nervosa	4 (3%)
Exercise induced amenorrhea	5/175 (3%)
Any amenorrhea > 3 months (other than while pregnant or nursing)	41/176 (23%)
Gastrointestinal Conditions
Celiac disease	11/174 (6%)
Ulcerative colitis	0
Crohn disease	0
Liver disease	5 (3%)
Primary biliary cirrhosis	0
Cystic fibrosis	0
Gaucher disease	0
Inflammatory Conditions
Lupus	1 (< 1%)
Rheumatoid arthritis	0
Other inflammatory arthritis	1 (< 1%)
Diabetes
Gestational diabetes	10/175 (6%)
Type 1 or type 2 diabetes	0
Any cancer diagnosis	6 (3%)
Multiple myeloma	0
Renal disease	1 (1%)
Renal stones	13 (7%)
Renal stone symptoms during pregnancy	5 (3%)
Alcohol and Smoking
Reported > 2 drinks/day of alcohol on average (past or current)	29 (16%)
Smoking
Current	6 (3%)
Ever	43 (24%)
Reported Medication Exposures
Glucocorticoid ever use	50 (28%)
Glucocorticoid use > 3 weeks	17 (10%)
Heparin/LMWH ever use	31/171 (18%)
Heparin/LMWH during pregnancy associated with PLO event	24 (14%)
DMPA ever use	11/172 (6%)
DMPA use prior to PLO pregnancy	10/175 (6%)
Progestin-only oral contraceptives after pregnancy	22/159 (14%)
Seizure medications/Antiepileptics	3/169 (2%)
Anti-cancer medications	1/170 (1%)

LMWH, low molecular weight heparin; DMPA, depot medroxyprogesterone acetate

Denominator 177 unless otherwise specified

^*Number of responders (denominator for categorical variables) is 177 unless otherwise indicated

Medication exposure

Medication exposures are also shown in Table 2. Oral steroid use was reported by 50 women (28%) with 17 (10%) participants reporting use > 3 weeks. Anticoagulation with heparin products during pregnancy was reported by 24 participants (14%), of whom the majority (n = 20) were treated with LMWH. DMPA use prior to pregnancy was reported by 10 participants (6%). Progestin only oral contraceptive use after pregnancy was reported by 22 (12%).

A substantial number of women reported use of low dose estrogen OCP (n = 28), SSRIs (n = 74) and prescription stimulants (n = 16). However, since timing of exposure in relation to onset of PLO fractures was not known, these exposures were not included in Table 2.

Predictors of disease severity (# of fractures):

Disease severity was defined as the total number of fractures during or after the first pregnancy associated with a PLO fracture. Number of fractures ranged from 1 to 12. Pre-specified clinical characteristics and conditions/medication exposures reported by 5 or more participants were tested as predictors of disease severity (Table 3).

Table 3.

Predictors of Disease Severity (Total Number of PLO Fractures) Predictors tested include prespecified clinical characteristics as well as conditions or medication exposures reported by 5 or more subjects. Parametric analyses are shown below

Categorical Variables	Number of Fractures (mean ± SD) in those with characteristic present vs not present
	Present	Not present	p
Family history of osteoporosis	4.7 ± 2.7	4.8 ± 2.7	0.77
History of childhood fracture(s)	4.5 ± 2.7	4.9 ± 2.7	0.36
Pregnancy characteristics related to PLO event
Primiparous at time of PLO event	5.0 ± 2.9	4.1 ± 2.3	0.04
Activity restriction/bedrest recommendation during pregnancy	4.5 ± 2.9	4.7 ± 2.7	0.67
PLO Fracture Site
Vertebral only	4.8 ± 2.6	4.1 ± 3.1	0.25
Hip only	1.8 ± 1.1	4.8 ± 2.7	0.02
PLO Fracture Timing
Fracture during pregnancy	3.5 ± 2.4	5.0 ± 2.7	0.004
Fracture during lactation	4.9 ± 2.7	4.1 ± 2.8	0.11
Continuous Variables	Correlation with Total No. of Fractures (Pearson r,p)
Age at first PLO fracture event	r = −0.11, p = 0.16
BMI	r = 0.07, p = 0.38
Age at menarche	r = −0.08, p = 0.31
Among those with fractures during lactation, months postpartum at time of event	r = −0.13, p = 0.08
Reported lumbar spine BMD T Score	r = −0.08, p = 0.330
Number of Fractures (mean ± SD) in those with characteristic present vs not present Or exposed vs not exposed
Reported Conditions/Medical History	Present	Not present	p
Hyperthyroidism (thyroid over-function)	5.2 ± 2.7	4.6 ± 2.7	0.64
Vitamin D deficiency (past or current)	4.8 ± 2.9	4.7 ± 2.5	0.80
Exercise induced amenorrhea	3.8 ± 3.1	4.7 ± 2.6	0.48
Any amenorrhea > 3 months (other than while pregnancy or nursing)	4.6 ± 3.0	4.7 ± 2.6	0.87
Celiac Disease	6.6 ± 3.3	4.5 ± 2.5	0.01
Liver disease	4.4 ± 2.1	4.7 ± 2.7	0.81
Gestational diabetes	4.4 ± 3.1	4.7 ± 2.6	0.72
Any cancer diagnosis	4.3 ± 2.7	4.7 ± 2.7	0.74
Renal stones	4.4 ± 2.7	4.8 ± 2.7	0.34
Past or current alcohol use > 2 drinks/day	4.1 ± 2.3	4.8 ± 2.8	0.17
Smoking (Ever)	4.7 ± 2.6	4.6 ± 2.7	0.80
Reported Medication Exposures	Exposed	Not Exposed	p
Oral Glucocorticoid ever use	4.8 ± 2.5	4.7 ± 2.8	0.70
Glucocorticoid use > 3 weeks	5.0 ± 3.2	4.7 ± 2.7	0.64
Heparin/LMWH ever use	5.6 ± 3.1	4.5 ± 2.6	0.07
Heparin use during pregnancy	5.7 ± 2.9	4.6 ± 2.6	0.03
DMPA use prior to PLO	3.2 ± 1.8	4.9 ± 2.6	0.07
DMPA ever use	3.8 ± 1.9	4.7 ± 2.8	0.28
Progestin-only oral contraceptives after pregnancy	4.3 ± 2.7	4.8 ± 2.7	0.48

Nonparametric results are presented in a Supplementary Table; bolded p values indicate statistical significance

Both having a fracture during pregnancy and having a hip fracture (which was more commonly reported during pregnancy) were associated with lower number of fractures. A fracture(s) that occurred during a primiparous delivery was associated with a higher number of fractures. Celiac disease was the only preexisting condition and heparin/LMWH use during pregnancy was the only medication exposure significantly related to disease severity.

Between groups comparisons

Between groups comparisons were conducted to define demographic, reproductive and fracture characteristic differences between those with and without celiac disease and with or without heparin/LMWH exposure (Table 4). Participants reporting celiac disease had significantly more fractures than those without celiac disease (6.6 ± 3.3 vs 4.5 ± 2.60; p = 0.01). Other clinical characteristics did not differ. Significance was maintained using non-parametric analyses (data not shown).

Table 4.

Between group comparisons related to conditions/medication exposures significantly related to disease severity

	Reported celiac disease diagnosis			Heparin/LMWH exposure during the pregnancy associated with the PLO event
	Celiac disease (n = 11)	No Celiacdisease (n = 161)	p	Heparin/LMWH (n = 24)	No heparin/LMWH (n = 153)	p
Age at first PLO fracture	32.3 ± 5.0	33.2 ± 4.0	0.30	34.6 ± 3.4	32.0 ± 5.1	0.01
Total number of PLO fractures	6.6 ± 3.3	4.5 ± 2.60	0.01	5.8 ± 3.0	4.5 ± 2.6	0.03
BMI	21.3 ± 4.2	22.4 ± 4.7	0.23	23.7 ± 6.1	22.1 ± 4.4	0.12
Age of menarche	12.8 ± 1.5	13.0 ± 1.6	0.65	12.7 ± 1.5	13.0 ± 1.6	0.13
Family history of osteoporosis	6 (55%)	84 (51%)	0.88	14 (58%)	75 (49%)	0.51
History of childhood fracture(s)	5 (45%)	69 (42%)	1.0	9 (38%)	64 (42%)	0.87
Pregnancy characteristics related to PLO event
Primiparous at time of PLO event	8 (73%)	108 (66%)	0.75	16 (64%)	100 (66%)	1.00
Activity restriction/bedrest recommendation during pregnancy	2 (18%)	26 (16%)	0.69	8 (32%)	23 (15%)	0.04
PLO Fracture Site
Vertebral only	11 (100%)	139 (85%)	0.40	18 (75%)	133 (87%)	0.13
Hip only	0	5 (3%)	0.56	1 (4%)	4 (3%)	0.52
PLO Fracture Timing
Fracture during pregnancy	0	31 (19%)	0.22	7 (28%)	24 (16%)	0.15
Fracture during lactation	11 (100%)	127 (77%)	0.12	14 (58%)	124 (80%)	0.02
Months postpartum at time of event	2.1 ± 1.5	2.0 ± 2.0	0.85	0.9 ± 0.9	2.1 ± 2.0	0.001
Reported BMD T score at the spine	−3.7 ± 0.7	−3.3 ± 1.8	0.41	−3.4 ± 1.0	−3.4 ± 1.2	0.95

Bolded p values indicate statistical significance

Participants who received anticoagulation with heparin or LMWH during the pregnancy had significantly more fractures than those who did not (5.8 ± 2.9 vs 4.5 ± 2.6 fractures; p = 0.03). Findings were similar when conducted using non-parametric analyses (data not shown). Participants reporting heparin use were also older (p = 0.02) and were more likely to fracture during lactation than during pregnancy. In addition, their fractures occurred earlier, averaging within 1 month of delivery compared to > 2 months in participants who did not take heparin (p = 0.003).

Discussion

In this study, we present a detailed analysis of the clinical features and characteristics of a large group of women who sustained fractures temporally associated with pregnancy and lactation. To our knowledge, this is the largest study of this rare disorder published to date. Participants completed an online survey detailing their experiences, with a particular focus on a comprehensive description of the fractures, their location, timing, association with trauma, and delays in diagnosis of their condition. Participants reported an average of 4–5 fractures, the majority (92%) symptomatic vertebral fractures that occurred during lactation and an average of 2 months after delivery. For most women (67%), the onset of fractures was associated with their first pregnancy. Unlike prior studies of PLO, we did not exclude patients with secondary causes of osteoporosis to clarify the heterogenous nature of medical history associated with this condition. Primiparity, heparin exposure during pregnancy and history of celiac disease were associated with more severe disease. These results provide important data that will advance our understanding of potential etiologies and mechanisms for this rare condition.

Our data are consistent in several respects with results of three previously published studies of PLO, including 29, 52 and 102 affected women [1, 5, 7]. Similar to our results, mean age at presentation was 29–34 years, and the majority presented with symptomatic vertebral fracture (> 75%) around a first full-term pregnancy (70–72%) [1, 5, 7]. Also similar to our findings, Hadji, et al. reported a mean of four fractures at presentation, with the great majority (95%) of fracture(s) diagnosed around 3 months post-partum [7]. Timing of presentation may have differed in the paper by Dunne, et al., who reported an onset of symptoms postpartum in only 48%, while the rest had symptom onset during pregnancy or delivery [1]. Similar to our study, prior studies have reported quite low spine BMD. Dunne et al. [1] reported a mean spine BMD Z score −2.8 and Laroche et al. [5] reported a median spine T score of −3.4. In a cohort of 52 patients, Laroche et al. reported that two had celiac disease and 3 had heparin exposure [5] – lower percentages than this cohort. Hadji et al. excluded women with known secondary causes present before pregnancy—but did report a similar percent with activity restriction (25.5% vs 18% in our study) during pregnancy [7].

Prior PLO studies differ from ours in terms of characteristics of the population included. We defined PLO based on fracture history, in contrast to a prior study [7] including some women without fracture history. We also included women with secondary causes of osteoporosis and medication exposures associated with bone loss, thus permitting us to study diseases and medication exposures that could potentially lead to fractures during pregnancy and lactation. In contrast, Dunne et al. [1] excluded women with secondary causes of bone loss that we included (celiac disease, use of anti-epileptics, anticoagulants or glucocorticoids). Hadji et al. [7] included women who received heparin and glucocorticoids during pregnancy, but excluded women with other disorders or pre-pregnancy medication exposures known to affect bone metabolism. Our study was larger and more diverse geographically than prior studies.

Our participants who had hip fractures can be compared to a cohort study of 33 women with pregnancy associated transient osteoporosis of the hip (TOH), defined as bone marrow edema in one or both femoral heads on MRI [11]. In this study [11], 12% had frank hip fracture. Nomenclature related to stress or insufficiency fractures of the femur during pregnancy has changed over time. The term “transient osteoporosis of the hip” was previously used to describe this condition, based on radiolucency seen on radiographs that were originally used for the diagnosis [12]. With the increasing use of MRI-based imaging, bone marrow edema findings can now lead to a clarified diagnosis of insufficiency fracture of the femur. We are classifying both marrow edema and frank hip fracture presentation as a hip fracture presentation of PLO [11].

Similar to our findings in PLO patients presenting with hip fracture, 30% of women included in the TOH study [11] had BMI > 26 kg/m² and commonly received recommendation for activity restriction or immobilization during pregnancy. Notably, it is unclear if BMI assessed at presentation also included those with BMI assessed while pregnant, and recommendations for immobilization may have occurred after the onset of TOH/hip fracture.

Our broad fracture inclusion criteria allowed us to investigate relationships between fracture timing and fracture type for the first time. Vertebral fractures typically occurred during lactation. In contrast, hip fractures typically occurred during pregnancy. That vertebral fractures occurred predominantly during lactation may be related to the particularly profound effects of lactation on spine BMD, a skeletal site rich in trabecular bone. Prior studies of site-specific changes in bone mass during pregnancy and lactation have documented that more bone loss occurs at sites with a high preponderance of trabecular bone (e.g., the spine) during lactation than during pregnancy [2, 9, 13]. Additionally, studies examining BMD change at multiple sites over lactation have consistently documented that the spine sustains the largest BMD loss during this time [13]. Although these changes are temporary in normal women and BMD usually recovers, women with PLO may have low vertebral bone mass antedating pregnancy that makes them more vulnerable to vertebral fractures during the physiologic changes in bone mass during lactation. In contrast, hip fractures that occur predominantly during pregnancy may be related to weight gain-induced changes in loading and mechanical stress.

We investigated potential secondary causes/contributors that could be related to this bone fragility presentation. Most women were primiparous (66.7%). Over 50% reported family history of osteoporosis. Celiac disease (6.2%) and gestational DM (5.6%) were the most commonly reported medical conditions. The most common medication exposure during pregnancy was heparin, mostly LMWH (14%). Notably, 13% of women reported using progestin only contraception after pregnancy. Among potential secondary causes investigated, the largest numbers of responders reported heparin use (24, 14%), celiac disease (11, 6.2%) and nephrolithiasis (13, 7.3%) during pregnancy.

We conducted analyses to identify potential predictors of disease severity. Use of heparin products during pregnancy and history of celiac disease were associated with increased disease severity. Primiparity was also associated with increased disease severity. There were no significant differences in other clinical characteristics between primiparous and multiparous women. However, some differences were noted between those with and without heparin exposure and those with or without reported celiac disease.

Prolonged anticoagulation is used during pregnancy or postpartum for women at high risk of venous thromboembolism, because of a prior history, a prosthetic heart valve, atrial fibrillation, history of fetal loss and reduced ejection fraction. Low molecular weight heparin (LMWH) is used more commonly than unfractionated heparin. LMWH binds to antithrombin, inhibiting activated Factor X and decreasing thrombin, fibrin and clot formation. Fourteen percent of our cohort were exposed to heparin during pregnancy, 83% of whom used LMWH. This exposure was associated with more PLO fractures and an earlier occurrence of PLO fractures in relation to delivery (0.9 ± 0.9 months after delivery in the heparin/LMWH exposure group vs. 2.1 ± 2.0 months after delivery in the no heparin/LMWH exposure group). It is possible that the relationship between heparin product use and number/timing of fractures is confounded by presence of certain indications for heparin use (such as cardiac disease or activity restriction) that could relate to fracture risk. However, we also note that, while activity restriction was more common in the heparin/LMWH exposure group, activity restriction/bed rest during pregnancy did not independently relate to disease severity.

Rodent studies investigating heparin effect on bone have shown decreased bone formation and inhibition of osteoprotegerin action, which in turn leads to increased bone resorption. The effects were greater with unfractionated heparin (UFH) compared to LMWH, though present in both [14–16]. In pregnant women, UFH use has been associated with bone loss [17, 18] and vertebral fractures in 2% [19]. Although LMWH is generally considered safer for the skeleton than UFH, its effects on bone density and fracture risk remain controversial. Significant bone loss has been reported in pregnant women treated with LMWH [20, 21]. Bone loss appears to be similar in comparison to untreated pregnant women [21], and studies have documented no osteoporotic/vertebral fractures in these cohorts [20–22]. Despite this reassuring data, our participants who used heparin products (mostly LMWH) during pregnancy had significantly more fractures than those without anticoagulation use. We have previously shown that bone formation rate (BFR) is significantly lower in women with PLO than in premenopausal women with idiopathic osteoporosis unrelated to pregnancy and in healthy controls, highlighting the potential mechanistic importance of bone formation or osteoblast defects [23]. We hypothesize that PLO women with underlying low BFR may be uniquely sensitive to effects of heparin products to decrease bone formation, increase bone resorption and uncouple bone remodeling.

Celiac disease is an autoimmune disorder characterized by small intestinal inflammation in genetically susceptible individuals triggered by gluten exposure [24]. One of the most common extraintestinal manifestations of celiac disease is low BMD and increased risk of fracture [25, 26]. In a meta-analysis of 16 studies, patients with celiac disease were found to have 30% and 69% increased risk of any fracture and hip fracture, respectively [27]. The pathophysiological mechanisms underlying bone loss in celiac disease remain incompletely understood. Hypothesized mechanisms include calcium malabsorption leading to secondary hyperparathyroidism and bone resorption [28, 29], changes in inflammatory cytokines leading to increased osteoclastogenesis [28, 30], hypogonadism, weight loss and malnutrition. In this study, celiac disease diagnosis was reported by 6% and was associated with higher number of fractures (disease severity) but not associated with other clinical factors.

Gastrointestinal absorption of calcium plays an important part in the calcium metabolism changes associated with normal pregnancy. During normal pregnancy, active 1,25 dihydroxy vitamin D levels rise substantially, mainly due to placental production, leading to a physiologic adaptive calcium hyperabsorption during pregnancy [13]. We hypothesize that women with malabsorptive disorders, such as celiac disease, may be at higher risk for fractures around pregnancy as they may be unable to increase calcium absorption during pregnancy, thus predisposing them to negative calcium balance earlier in pregnancy and contributing to fracture risk in the context of the skeletal stress of pregnancy and lactation.

This study has several limitations. Data collection was obtained through questionnaires and all data is self-reported. Fractures were not confirmed radiographically. This study lacks a control group as all participants were recruited based on history of fracture. Despite the lack of a control group, some variables such as vitamin D deficiency are comparable to the general population. Our study found that 43% of participants reported vitamin D deficiency which is comparable to prior studies assessing prevalence of vitamin D deficiency in pregnant women. A study in white and black pregnant women in northeast USA found 60% of white women and 51% of black women were vitamin D insufficient (37.5 – 80 nmol/L) at 4–21 weeks’ gestation and above 40% remained insufficient at 37–42 weeks of gestation [31]. A meta-analysis in healthy pregnant women in India found a prevalence of 30% Vitamin D deficiency with higher prevalence in other developing countries [32].

With respect to fracture timing, we were not able to clarify whether timing of fracture was based on symptom onset or radiographic diagnosis. Pain symptoms may not have been temporally associated with specific traumatic events. Bone density reports were not available. Patients reporting vitamin D deficiency did not base diagnosis on a standardized lab cutoff for deficiency. In our patient cohort, it is important to note that subjects self-identified as having celiac disease. Confirmation via pathology or serologies (anti-tissue transglutaminase antibodies) was not available. It is possible that some reporting celiac disease would not meet serologic criteria for the diagnosis. It is also possible that some women not reporting celiac disease could have undiagnosed celiac disease.

In conclusion, this is the largest study of pregnancy and lactation associated osteoporosis yet reported. We found that the average number of fractures, type of and timing of fractures are similar to previous smaller studies. Both the size of this study and the inclusion of women with a broad range of clinical and fracture characteristics provide novel information about the characteristics and potential risk factors for PLO. In addition, several factors, including primiparity, exposure to heparin products and celiac disease were associated with greater severity, as evidenced by higher number of fractures. Future studies are necessary to define mechanisms by which heparin exposure or celiac disease impact the clinical presentation of PLO. This data advances our understanding of potential etiologies for this rare condition and highlights the importance of developing a mechanistic understanding towards treatment approaches.

Supplementary Material

Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s00198-023-06793-9.

Data Availability

The data used to support the findings of the study are included within the article.