Possible role of bone turnover markers in the diagnosis of adult hypophosphatasia

Abstract

Hypophosphatasia (HPP) is a rare disorder of the bone metabolism, characterized by genetically determined low alkaline phosphatase (ALP) activity. Low ALP may also be observed in some common causes of bone fragility, such as in osteoporosis treated with antiresorptive drugs. This study aimed to verify whether differences in bone turnover markers (BTMs) could help differentiate adult patients with HPP from those with osteoporosis undergoing antiresorptive treatment. In this multicenter study, we enrolled 23 adult patients with a diagnosis of HPP and compared them with 46 osteoporotic subjects previously treated with zoledronic acid or denosumab. BTMs such as CTX, N-terminal propeptide of type I procollagen (P1NP), total ALP, and bone ALP (bALP) were measured, and ratios between BTMs were also calculated. Considering that the control group included only females, in the primary analysis we compared their characteristics with that of the 16 female patients with HPP. Both individual BTMs (CTX and P1NP) and 4 BTM ratios (ALP/P1NP, bALP/P1NP, ALP/CTX, and bALP/CTX) showed satisfactory discriminatory power, outperforming ALP alone. P1NP, in particular, had an area under the curve (AUC) of 0.962 with a cut-off of 32 μg/L, while as for the BTMs ratios, the ALP/P1NP ratio had an AUC of 0.964 with a cut-off of 1.114. Similar results were confirmed when including male HPP patients, when adjusting for age and sex, and finally when performing a sensitivity analysis only in patients with ALP less than or equal to 32 U/L (ie, the median of the distribution of the entire population). In cases of low ALP and bone fragility, BTM and their ratios could help distinguish HPP patients from osteoporotic individuals treated with antiresorptive drugs, aiding in accurate diagnosis and reducing the risk of inappropriate treatment.

Introduction

Hypophosphatasia (HPP) is a rare metabolic bone disease, caused by inactivating mutations in the ALPL gene, which encodes for the tissue-nonspecific alkaline phosphatase (TNSALP or ALP),¹ a key enzyme in bone mineralization. Low ALP activity may be associated with impaired bone mineralization which may result in skeletal fragility and stress fractures.¹ Based on clinical, radiological, biochemical, and histopathological characteristics, HPP has an extraordinary range of severity, from stillbirth to forms with only dental complications.² Milder forms of HPP, especially in adults, are frequently misdiagnosed as other more common bone disorders, such as osteoporosis.^3,4 This diagnostic error could be harmful in patients with HPP as it could result in the use of antiresorptive drugs that can further impair bone mineralization,⁵ increasing the risk of fractures.⁶ The hallmark of HPP is low serum ALP. However, low ALP activity can be observed in many other conditions, such as endocrine disorders (hypoparathyroidism, hypothyroidism, hypercortisolism or glucocorticoid therapy), gastrointestinal and liver disorders (coeliac disease, inflammatory bowel diseases, Wilson’s disease), zinc or magnesium deficiency, malnutrition, and anorexia nervosa.^3,7 Furthermore, in the context of bone fragility, low ALP activity levels are frequently observed in patients treated with potent antiresorptive drugs (ie, i.v. bisphosphonates and denosumab), which decrease bone turnover.^8,9 To improve the diagnostic workup of HPP patients, many diagnostic algorithms for the classification of adult patients with low ALP have been proposed.^7,10,11 These imply the use of ALPL gene analysis or the assay of additional biochemical markers, such as serum pyridoxal-5′-phosphate (PLP, a biologically active form of vitamin B6) or urinary phosphoethanolamine (PEA), both of which are metabolites that can accumulate in HPP. These approaches appear to be effective in increasing the detection rate of HPP among patients with low ALP activity,⁷ but assays can be expensive and not readily available in many laboratories. In contrast, there is no evidence regarding the use of the bone turnover markers (BTMs), commonly employed in the clinical management of osteoporosis. Those most frequently used are either the product of osteoblasts, released during bone formation (eg, the aforementioned ALP and its skeletal isoform or bone ALP, and the N-terminal propeptide pro-collagen I or P1NP) or product of osteoclast-induced bone degradation, released during bone resorption (eg, CTX).¹²

López-Delgado et al. observed that in subjects with low ALP activity and no overt manifestations of HPP, CTX and P1NP were lower than in healthy controls with ALP within the reference range. Conversely, in subjects with low ALP activity, CTX and P1NP values were similar in those with ALPL mutation as compared to those without.¹³ Hepp et al. observed that in patients with confirmed HPP, bone ALP (bALP) and P1NP, but not CTX, were lower than in healthy controls.¹⁴ Finally, Desborough et al. observed that in HPP patients, compared to subjects with low BMD from other causes, ALP and bALP levels were lower, CTX levels were higher, and no differences were observed in P1NP levels. It should be noted that in the low BMD group, 61.9% of patients were treated with bisphosphonates.¹⁵ Since data available on the role of BTMs in the diagnosis of adult HPP are inconclusive, the purpose of this study was to evaluate whether the use of different BTMs may increase their diagnostic power in this setting. In addition, considering that serum ALP activity levels are related to the presence of an ALPL gene mutation in HPP patients and not the expression of the bone remodeling activity, we wanted to evaluate if the ratio between serum ALP and serum CTX and P1NP values, not affected by the genetic alteration, may further contribute to the diagnostic process of these patients. In particular, we wanted to compare HPP patients with osteoporotic subjects treated with potent antiresorptive drugs, frequently characterized by low values of ALP activity and bone fragility.

Materials and methods

Patients

This was a multicenter study in which we enrolled patients aged ≥18 yr, previously diagnosed with HPP, and followed from 2014-2022 in the outpatient clinic of 5 hospitals in Italy (Azienda Ospedaliera Universitaria Integrata, Verona; Azienda Ospedale-Università, Padua; University Hospital Policlinico Umberto I, Rome; IRCCS “Casa Sollievo della Sofferenza”, San Giovanni Rotondo; ASST-Pini-CTO, Milan). The diagnosis of HPP was based on the presence of at least 2 plasma ALP activity measurements below the age- and sex-adjusted reference range¹⁶ and more than one of the following: (1) pathogenic or likely pathogenic ALPL gene variant; (2) serum PLP values above the upper limit of normal; (3) history of low-trauma fractures (especially of the long-bones) and chronic musculoskeletal pain. From an initial cohort of 31 HPP subjects, 8 patients were not included because of glucocorticoid use due to chronic inflammatory conditions (2 patients), previous treatment with bisphosphonates (1 patient), previous treatment with asfotase alfa (1 patient), chronic kidney disease (CKD) stage 3 (1 patient), pregnancy (1 patient), hypothyroidism (1 patient), or declining to participate in the study (1 patient). To increase the statistical power of the study, we compared each of the 23 remaining HPP patients with 2 osteoporotic control subjects. These were consecutively enrolled in the 5 outpatient clinics if they met the following inclusion criteria: (1) age ≥ 18 yr; (2) history of bone fragility, defined by the presence of at least one fragility fracture and low BMD (hip and/or lumbar spine T-score ≤ −2.0); (3) plasma ALP below the age- and sex-adjusted reference range¹⁶, documented within 12 mo after the use of i.v. zoledronic acid 5 mg or within 6 mo after s.c. denosumab 60 mg; (4) plasma ALP within the age- and sex-adjusted reference range¹⁶ before initiating the aforementioned treatment; (5) no familial history of HPP. Exclusion criteria for controls were: presence of secondary causes of bone fragility or other metabolic bone disorders (eg, primary hyperparathyroidism, hypercortisolism or glucocorticoid therapy, coeliac disease, inflammatory bowel diseases, malnutrition, anorexia nervosa, hypoparathyroidism, Paget's disease of bone), CKD stage 3–5 (estimated glomerular filtration rate [eGFR] < 60 mL/min/1.73 m², according to Chronic Kidney Disease Epidemiology Collaboration equation), hypothyroidism, zinc or magnesium deficiency, Wilson’s disease, primary bone cancer or bone metastasis due to other neoplasm, pregnancy, or breastfeeding. Patients and controls had cholecalciferol treatment to correct hypovitaminosis D, when present.

Data collection

Medical records were reviewed and demographic, clinical, radiological, and laboratory data were collected. These included the following: age, sex, height, weight, BMI, past medical history, presence of fragility fractures, ongoing medical treatment, laboratory tests (serum creatinine, serum and urinary calcium, serum and urinary phosphate, PTH, 25-OH-vitamin D, total ALP), BMD T-scores (measured at single center level using dual-energy X-ray absorptiometry machines), and genetic analysis in HHP subjects, when available. After their inclusion, all HPP and control patients underwent fasting blood sampling for centralized measurement of ALP, bALP, CTX, P1NP, PLP (HPP only), and 25-OH-vitamin D.

Biochemical analysis

Fasting serum calcium, phosphate, creatinine, and PTH, as well as urine calcium and phosphate, were retrieved from local medical records. PTH was measured by a second-generation assay in 4 out of 5 centers, while a third-generation assay (automated immunochemiluminescent method, Liaison XL, DiaSorin, Saluggia, Italy; reference range 6.5-36.8 ng/L) was used at 1 center. Fasting ALP activity, measured prior to starting antiresorptive treatment in osteoporotic patients and in all HPP subjects, was also retrieved from local medical records. Morning fasting blood samples were obtained from all patients for the following centralized assays. Serum 25-OH-vitamin D was measured using an automated immunochemiluminescent method with a LIAISON XL Assay (DiaSorin Inc., Stillwater, MN, United States). ALP activity was measured in lithium heparin plasma with an enzymatic-colorimetric method (Roche Diagnostics, Penzberg, Germany; reference range 46-122 U/L). CTX, P1NP, and bALP were assayed by an immunochemiluminescent method on an IDS-iSYS system (Pantec, Torino, Italy). Serum PLP was assayed by an HPLC method in HPP patients only; none of them was taking vitamin B6 supplements. The following ratios were then calculated for all subjects: ALP/P1NP, bALP/P1NP, ALP/CTX, bALP/CTX, P1NP/CTX, and bALP/ALP.

Statistical analysis

Normally distributed continuous variables were expressed as mean​​ ± SD. Non-normally distributed continuous variables were expressed as median and interquartile range (IQR). The T-test for independent observations was used to compare normally distributed continuous variables, and the Mann–Whitney test for non-normally distributed continuous variables between 2 groups. The chi-square test was used to compare categorical variables. The calculation of the receiver operating characteristic (ROC) curves and related areas under the curve (AUC) were used to evaluate the diagnostic power of BTMs and their ratios, alone and combined with age and sex. The optimal cut-off value for each BTM or ratio was determined by calculating the Youden index. Sensitivity, specificity, and positive and negative likelihood ratios were calculated for each ROC curve. To account for differences in sex and ALP distribution between cases and controls, sensitivity analyses in females and in those with ALP ≤32 U/L (corresponding to the median value of this biomarker in the whole cohort of patients and controls) were performed. Comparison between different AUCs was performed with the DeLong method.¹⁷ Results were considered significant when the p-value was <0.05. Statistical analysis was performed using the software Statistical Package for the Social Sciences (SPSS Statistics for Windows, Version 27.0. Armonk, NY: IBM Corp).

Results

Patient characteristics

A total of 69 patients were included. Of these, 23 were HPP subjects and 46 were affected by primary osteoporosis and on treatment with antiresorptive drugs. Inclusion characteristics for patients only with HPP are reported in Figure 1 and further detailed in Table S1. In all of these patients, the disease was diagnosed in adulthood, and none had received osteoporotic treatments. Among the control patients, 5 had been treated with zoledronic acid (median treatment duration of 24 mo, with a median time since the last infusion of 7 mo) and 41 with denosumab (median treatment duration of 18 mo, with a median time since the last injection of 4 mo). Median ALP activity levels prior to the initiation of antiresorptive drugs in these subjects were 67 U/L (IQR 58-86).

Figure 1.

Characteristics of patients with HPP according to inclusion and exclusion criteria. Abbreviations: ALP, alkaline phosphatase; CKD, chronic kidney disease; HPP, hypophosphatasia.

In the cohort thus identified, 16 out of 23 HPP patients were female (72%), while only females were included in the control group. Given this different distribution of sexes between the 2 groups, we initially analyzed only the female population, whose characteristics are reported in Table 1. Nine out of 16 female patients with HPP were postmenopausal, as were all controls. Among HPP patients, we did not observe any significant differences in any BTM between premenopausal and postmenopausal women (p>0.05 for all comparisons). Regarding fragility fractures, patients with HPP had a rate of prevalent femoral fractures of 12.5%, other long bones fractures of 68.8%, metatarsal fractures of 25.0%, and vertebral fractures of 6.3%. In patients with primary osteoporosis, previous fracture rates for hip, spine, and other long bone or major osteoporotic fractures were 6.5%, 60.2%, and 39.1%, respectively. In both groups, no fracture occurred in the year prior to enrollment in the study. In terms of BMD, female subjects with HPP had higher lumbar spine T-scores of −1.20 (IQR −2.30 to −0.30) compared to controls with −2.45 (IQR −3.05 to −2.10, p<0.001). In contrast, no significant differences were observed for total hip and femoral neck T-scores.

Table 1.

Female patient characteristics.

Characteristic	Hypophosphatasia   (n = 16)	Controls   (n = 46)	p-value
General
Age (yr)	50.9 ± 13.4	65.1 ± 8.6	<0.001
Height (cm)	163.1 ± 8.3	158.2 ± 4.1	0.479
Weight (Kg)	59.3 ± 12.1	58.1 ± 12.2	0.785
BMI (Kg/m2)	23.4 ± 4.8	23.3 ± 4.9	0.979
Laboratory tests
Calcium (mg/dL)	9.4 (9.0–9.6)	9.3 (9.0–9.6)	0.782
Phosphate (mg/dL)	3.8 (2.9–4.4)	3.5 (3.1–3.8)	0.197
Creatinine (mg/dL)	0.70 (0.62–0.86)	0.90 (0.87–0.97)	0.003
eGFR (mL/min/1.73 m2)	86 (75–101)	94 (87–99)	0.356
25-OH-vitamin D (nmol/L)	81.1 (54.5–97.0)	90.0 (75.0–112.5)	0.061
PTH (2° generation assay, pg/mL)	49.0 (43.0–60.0)	51.4 (38.7–66.0)	0.789
PTH (3° generation assay, ng/L)	21.2 (15.0–26.0)	26.3 (14.9–34.5)	0.831
Urinary calcium (mg/24 h)	159 (112–223)	187 (147–216)	0.305
Urinary phosphate (mg/24 h)	539 (480–600)	640 (471–703)	0.831
Bone turnover markers
CTX (pg/mL)	299 (149–363)	33 (30–120)	<0.001
ALP (U/L)	25 (21–32)	37 (30–41)	0.001
bALP (μg/L)	4.7 (3.2–6.4)	6.1 (3.9–7.3)	0.085
P1NP (μg/L)	61 (42–77)	13 (9–20)	<0.001
Bone turnover markers ratios
ALP/P1NP ratio	0.60 ± 0.53	2.66 ± 1.07	<0.001
bALP/P1NP ratio	0.10 ± 0.07	0.47 ± 0.30	<0.001
ALP/CTX ratio	0.15 ± 0.17	0.82 ± 0.48	<0.001
bALP/CTX ratio	0.03 ± 0.02	0.13 ± 0.08	<0.001
P1NP/CTX ratio	0.35 ± 0.37	0.35 ± 0.24	0.975
bALP/ALP ratio	0.20 ± 0.07	0.18 ± 0.09	0.474

Data are expressed as mean ± SD or median and interquartile range, as appropriate. Abbreviations: ALP, alkaline phosphatase, bALP, bone alkaline phosphatase; eGFR, estimated glomerular filtration rate; P1NP, procollagen type 1 N-terminal propeptide.

Bone turnover markers and ratios

Values of BTMs in female patients are reported in Table 1, and their distributions are shown in Figure S1. Patients with HPP were characterized by significantly lower ALP activity (median 25 U/L, IQR 21–32 U/L) than controls (median 37 U/L, IQR 30–41 U/L); this difference was not evident for bALP. In contrast, patients with primary osteoporosis on antiresorptive treatment had significantly lower levels of serum P1NP (p<0.001) and CTX (p<0.001) compared to HPP subjects. Four out of the 6 BTM ratios (ALP/P1NP, bALP/P1NP, ALP/CTX, and bALP/CTX, p<0.001 for all comparisons) were significantly lower in patients with HPP than in osteoporotic subjects, while no differences were found for the P1NP/CTX and bALP/ALP ratio. The distributions of BTM ratios are shown in Figure S2.

ROC curve analysis

Receiver operating characteristic (ROC) analysis was then performed to evaluate the discriminatory power of individual BTMs and their ratios in distinguishing female HPP patients from osteoporotic subjects with low ALP activity. As shown in Figure 2 and in Table 2, ALP, CTX, and P1NP, but not bALP, demonstrated discriminatory power between the 2 conditions. With regard to BTM ratios, P1NP/CTX and bALP/ALP ratios did not demonstrate any discriminatory power between the 2 groups, while the performance of all other evaluated ratios was at least satisfactory. The best cut-offs for each BTM or ratio, defined by the Youden index method, as well as sensitivity, specificity, positive and negative likelihood ratios are also reported in Table 2.

Figure 2.

ROC curves of bone turnover markers (A) and their ratios (B) in female patients. Abbreviations: ALP, alkaline phosphatase; bALP, bone alkaline phosphatase; P1NP, procollagen type 1 N-terminal propeptide; ROC, receiver operating characteristic.

Table 2.

Area under the curve (AUC) and related indexes for ROC curves of different BTMs and their ratios in female patients.

Test result variables	AUC	SE	p-value	95% CI	Best   cut-off	Sensitivity	1 – Specificity	+LR	−LR	Youden index
ALP	0.763	0.066	0.001	0.634–0.892	35	0.596	0.118	5.1	0.458	0.478
bALP	0.556	0.092	0.494	0.377–0.736	-	-	-	-	-	-
CTX	0.894	0.040	<0.001	0.815–0.972	67	0.941	0.340	2.8	0.089	0.601
P1NP	0.962	0.041	<0.001	0.908–1.000	32	0.941	0.021	44.8	0.060	0.920
ALP/P1NP ratio	0.964	0.027	<0.001	0.911–1.000	1.114	0.979	0.059	16.6	0.022	0.920
bALP/P1NP ratio	0.956	0.023	<0.001	0.910–1.000	0.255	0.787	0.001	787	0.213	0.786
ALP/CTX ratio	0.938	0.033	<0.001	0.873–1.000	0.204	0.915	0.235	3.9	0.111	0.680
bALP/CTX ratio	0.915	0.042	<0.001	0.833–1.000	0.071	0.723	0.059	12.3	0.294	0.665
P1NP/CTX ratio	0.552	0.085	0.538	0.386–0.718	-	-	-	-	-	-
bALP/ALP	0.616	0.075	0.160	0.469–0.762	-	-	-	-	-	-

Abbreviations: ALP, alkaline phosphatase; AUC, area under the curve; bALP, bone alkaline phosphatase; BTMs, bone turnover markers; P1NP, procollagen type 1 N-terminal propeptide; ROC, receiver operating characteristic; +LR, positive likelihood ratio; −LR, negative likelihood ratio.

We then compared the AUCs of all significant ROC curves. Initially, we examined the AUCs of the individual BTMs (ie, CTX and P1NP) and BTM ratios (ie, ALP/P1NP, bALP/P1NP, ALP/CTX, and bALP/CTX) against that of ALP, revealing that all outperformed ALP (p<0.05 for all comparisons). Considering P1NP as the BTM with the best diagnostic performance based on AUC, we further assessed its differences compared to CTX alone and BTM ratios. However, we did not observe any significant difference between the AUC of P1NP and the AUCs of CTX and BTM ratios (p>0.05 for all comparisons).

Furthermore, we decided to investigate whether the results observed in the female-only group were consistent when extending the analysis to a population that also included male HPP subjects. The characteristics of the patients included in this analysis are provided in Table S2. In keeping with the results obtained for female-only group, it was confirmed even in this analysis that ALP, CTX, P1NP, and 4 BTM ratios (ALP/P1NP, bALP/P1NP, ALP/CTX, and bALP/CTX) had a discriminatory power in differentiating between HPP and controls (Table S3 and Figure S3).

Given the fact that in addition to sex, age also significantly differed between cases and controls, we assessed to what extent these 2 variables modified the discriminatory power of the BTMs and the aforementioned ratios for distinguishing patients with HPP from those without HPP. As shown in Table S4 and Figure S4, we observed that adjusting for age and sex increased the discriminatory power of P1NP and CTX by 4.2% and 10.3%, of ALP and bALP by 12.0% and 25.1%, and of the ALP/P1NP and bALP/P1NP ratios by 2.0% and 3.2%, respectively. This indicates that age and sex can further refine the diagnostic accuracy of these biomarkers in identifying patients with HPP.

Finally, since we observed a difference in ALP activity between the HPP patients and controls, we carried out a sub-analysis including only male and female subjects with ALP activity levels ≤32 U/L (corresponding to the median value of the biomarker in the whole group of patients and controls), i.e. 17 subjects affected by HPP and 18 controls. ALP activity levels in HPP patients were 24 U/L (IQR 21–28), while in controls were 28 U/L (IQR 24–30), showing no significant difference (p=0.123). However, even in this analysis it was confirmed that CTX, P1NP, and 4 ratios had high discriminatory power in distinguishing between HPP and controls with the following AUCs (CTX 0.885, P1NP 0.969, ALP/P1NP 1.000, bALP/P1NP 0.964, ALP/CTX 0.924, bALP/CTX 0.915, p<0.001 for all comparisons), as shown in Figure S5.

Discussion

This study demonstrates that the use of BTMs and their ratios may help distinguish HPP patients from other subjects with low ALP activity and bone fragility, such as osteoporotic patients undergoing antiresorptive therapy. This observation is important because HPP is characterized by a variable, nonspecific clinical and radiological presentation that, in many cases, may resemble signs and symptoms observed in osteoporosis. Although the confirmation of an HPP diagnosis almost always requires genetic analysis of ALPL variants, and this test is becoming more widely available, it may be useful to have simple tools that can increase diagnostic suspicion and guide the subsequent diagnostic process, particularly in settings with limited diagnostic resources. Other laboratory tests (eg, PLP and urinary PEA) are used to aid in the diagnosis of HPP,⁷ but these assays are not easily available and can be costly. In contrast, the evaluation of BTMs is performed routinely in laboratories specialized in metabolic bone diseases. BTMs help clinicians in evaluating fracture risk in osteoporotic patients,¹⁸ estimating adherence to anti-osteoporosis treatment and may contribute to predict therapeutic efficacy, especially when antiresorptive treatments are used.¹⁹ These drugs exert marked anti-osteoclastic activity, and suppress bone resorption marker levels, such as serum CTX.^8,9 However, because of the coupling phenomenon, they are known to influence the entire bone remodeling process, inducing a fall in bone formation markers, such as ALP and P1NP.^8,9 It is worth noting that values of ALP below the age- and sex-adjusted reference range have been reported in some patients treated with potent antiresorptive.^3,15 On the other hand, how HPP may influence levels of BTMs different from ALP, such as P1NP and CTX, remains to be elucidated. Data currently available are scanty and difficult to interpret because they frequently refer to non-homogeneous populations (ie, ascertained vs. not ascertained HPP, low BMD-treated subjects vs. healthy untreated controls).^13,14 However, some studies have already reported that serum CTX and P1NP may be higher in HPP patients compared to subjects with low ALP activity, such as antiresorptive-treated osteoporotic patients.¹⁵ In the present study, we compared HPP patients with a population of osteoporotic subjects with clinically proven bone fragility, normal ALP activity before treatment and ALP activity below age- and sex-adjusted reference range¹⁶ after the therapeutic use of zoledronic acid or denosumab. We found that ALP activity was low in both patients and controls, even if it was still lower in HPP subjects. We also observed that while serum bALP values did not differ between the 2 groups, P1NP and CTX serum levels were markedly lower in patients treated with antiresorptives than in HPP subjects. In our view, the trend of BTMs in the control group may be explained by the coupling phenomenon in the bone remodeling process reached over time when using bisphosphonates and denosumab. On the contrary, in HPP patients there is a dissociation between the genetically determined low ALP activity and serum P1NP and CTX values, which are not affected by the ALPL gene mutation. In fact, we observed that both P1NP and CTX yielded a good discriminatory power between the 2 conditions, which is superior to that of ALP. Moreover, together with individual BTMs, to better highlight the different trends among them, we also calculated their ratios. Although other methods for the combined assessment of 2 BTMs have been described, such as T-scores,²⁰ we preferred the use of their ratios. We believe this approach is more applicable in clinical practice, since it does not require knowledge of the biomarker distribution in the population. However, since there is no standardization in BTM assays, the cut-offs derived from this study may slightly vary depending on the specific assay used. Four of the 6 BTM ratios (excluding P1NP/CTX ratio and bALP/ALP) had a diagnostic performance that was at least good or even excellent in distinguishing HPP subjects from controls. In particular, we identified specific cut-off values for BTMs and their ratios, with a sensitivity ranging from 72.3-97.9% and a specificity ranging from 66.0-99.9% when analyzing both sexes. For example, according to our findings, a female patient exhibiting low ALP activity and a P1NP value of 32 or higher or an ALP/P1NP ratio of 1.114 or lower is likely to be affected by HPP, with a positive likelihood ratio of 44.8 and 16.6, respectively. This means that by using laboratory tests commonly available, a very robust suspicion of HPP may be proposed, as an additional and easy-to obtain tool in the diagnostic workup of patients with low ALP activity and bone fragility. It is interesting to note that the P1NP/CTX ratio did not show any discriminatory power in terms of AUC. This can be explained by the fact that BTM ratios not involving ALP and bALP do not have any reason to be different between HPP patients and subjects treated for bone fragility, in which bone formation and resorption are supposed to be possibly coupled. Due to this, the fact that P1NP/CTX ratio was almost identical between HPP patients and controls may further represent a useful diagnostic element when looking to patients with low ALP activity and bone fragility.

This study has several limitations. The sample size was relatively small. However, it should be noted that HPP is a rare disease. Moreover, to reduce the possibility of bias, we used stringent criteria for its diagnosis. Accordingly, 8 out of 31 patients were excluded because of concomitant clinical conditions possibly acting as confounding factors. In the control group we only included patients with low ALP activity secondary to the use of antiresorptive drugs. Our choice was based on the fact that the above-mentioned subjects are those in which low ALP activity and a clinical pattern of bone fragility are more commonly associated. Indeed, among osteoporotic patients treated with antiresorptives for 3–5 yr, ALP activity levels were found below the reference range in 5% of those on alendronate, 25% of those on denosumab, and 33% of those on zoledronic acid.⁷ We are well aware that low ALP activity can be observed in many other conditions which can be encountered in clinical practice (eg, zinc deficiency, coeliac disease, glucocorticoid use).^3,7 However, with the exception of glucocorticoid-induced osteoporosis, these are generally underrepresented as a cause of low ALP activity. The role of BTMs in these conditions certainly deserves further investigation in future studies. In addition, to further increase the statistical power of the analyses, we selected 2 controls for each patient with HPP. In control patients, we did not perform genetic testing for ALPL variants to definitively rule out HPP. However, we included only those with normal ALP before starting antiresorptive treatment, which should indicate a low probability of HPP.

Considering that our control patients included only females, we first performed analysis in the female-only group of HPP patients, followed by analysis in both sexes, obtaining almost identical results in terms of the discriminatory power of BTM ratios. Moreover, the discriminatory power of the BTMs and their ratios for distinguishing patients and controls was further refined by taking into consideration not only sex but also age. It should be noted that the discriminatory power of CTX, P1NP, and the BTM ratios is less affected by age compared to that of ALP, and particularly bALP. This may suggest greater reliability of the former in distinguishing between patients and controls, but it also underscores the significant impact that age can have on the interpretation of some BTMs. The influence of age on BTMs could be attributed to factors beyond age-related modifications, such as physical activity levels, mobility impairments, the use of ambulatory assist devices, and hormonal status. We also acknowledge that menopausal status may indeed influence BTM levels. In our study, the proportion of postmenopausal women was indeed higher in the control group than in the HPP cohort. However, it is important to note that all patients in the control population were treated with antiresorptive agents, which effectively reduced both bone formation and resorption BTMs, thereby mitigating the potential impact of menopausal status on BTM levels in this group. Instead, the premenopausal women in the HPP cohort were younger than the postmenopausal women, so menopausal status in the HPP cohort may have influenced BTM values in an age-dependent manner.

Finally, even if, by definition, our whole patient population had ALP activity below the age- and sex-adjusted reference range, ALP was lower in HPP subjects than in controls. This may suggest that the severity of the ALP reduction might serve as an additional clue in distinguishing HPP subjects from osteoporotic ones undergoing antiresorptive treatments. To mitigate potential bias arising from this difference, we conducted a sensitivity analysis including only subjects with lower ALP (ie, below the median value: 32 U/L), to obtain groups with similar ALP activity levels. Once again, the discriminatory power of the BTMs and their ratios was confirmed even in this subset. In addition, it should be noted that, as reported by others,⁴ these 2 groups had overlapping serum bALP levels along with almost similar clinical presentations in terms of bone fragility, with this contributing to possible errors in the diagnosis of these patients.

In conclusion, these findings highlight the possible relevant role that BTMs, along with clinical and radiological characteristics, may play in the initial phase of the HPP diagnostic workup, thus hopefully leading to an improvement in diagnostic delay, and reducing the risk of misdiagnosis and potential harm from inappropriate treatment in these patients.

Author contributions

Francesco Bertoldo (Conceptualization, Formal analysis, Investigation, Writing—original draft, Writing—review & editing), Giovanni Tripepi (Conceptualization, Formal analysis, Writing—review & editing), Martina Zaninotto (Conceptualization, Investigation, Writing—review & editing), Mario Plebani (Writing—review & editing), Alfredo Scillitani (Conceptualization, Investigation), Massimo Varenna (Conceptualization, Investigation), Chiara Crotti (Data curation), Cristiana Cipriani (Data curation), Jessica Pepe (Data curation), Salvatore Minisola (Conceptualization, Investigation), Flavia Pugliese (Data curation), Vito Guarnieri (Data curation), Valeria Baffa (Data curation), Marco Onofrio Torres (Data curation), Francesca Zanchetta (Data curation), Maria Fusaro (Formal analysis), Maurizio Rossini (Writing—review & editing), Maria Luisa Brandi (Writing—review & editing), Colin Egan (Writing—review & editing), Paolo Simioni (Writing—review & editing), Gaetano Paride Arcidiacono (Conceptualization, Formal analysis, Investigation, Writing—original draft, Writing—review & editing), Stefania Sella (Conceptualization, Investigation, Writing—original draft, Writing—review & editing), and Sandro Giannini (Conceptualization, Formal analysis, Investigation, Writing—original draft, Writing—review & editing)

Funding

None declared.

Conflicts of interest

None declared.

Data availability

All datasets generated during and/or analyzed during the current study are not publicly available but may be available from the corresponding author on reasonable request.

Ethics approval and informed consent

This study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki of 1975, as revised in 2000. The study was approved by the local Ethics Committee (CET-ACEV n° 412n/AO/23). All participants provided informed consent prior to their inclusion in the study.