Marrow Cell Transplantation for Infantile Hypophosphatasia


  • Presented in part at the 20th Annual Meeting of the American Society for Bone and Mineral Research, December 1-6, 1998, San Francisco, CA, USA.

    The authors have no conflict of interest.


An 8-month-old girl who seemed certain to die from the infantile form of hypophosphatasia, an inborn error of metabolism characterized by deficient activity of the tissue-nonspecific isoenzyme of alkaline phosphatase (TNSALP), underwent the first trial of bone marrow cell transplantation for this heritable type of rickets. After cytoreduction, she was given T-cell-depleted, haplo-identical marrow from her healthy sister. Chimerism in peripheral blood and bone marrow became 100% donor. Three months later, she was clinically improved, with considerable healing of rickets and generalized skeletal remineralization. However, 6 months post-transplantation, worsening skeletal disease recurred, with partial return of host hematopoiesis. At the age of 21 months, without additional chemotherapy or immunosuppressive treatment, she received a boost of donor marrow cells expanded ex vivo to enrich for stromal cells. Significant, prolonged clinical and radiographic improvement followed soon after. Nevertheless, biochemical features of hypophosphatasia have remained unchanged to date. Skeletal biopsy specimens were not performed. Now, at 6 years of age, she is intelligent and ambulatory but remains small. Among several hypotheses for our patient's survival and progress, the most plausible seems to be the transient and long-term engraftment of sufficient numbers of donor marrow mesenchymal cells, forming functional osteoblasts and perhaps chondrocytes, to ameliorate her skeletal disease.


HEMATOPOIETIC STEM CELL transplantation can cure malignant osteopetrosis,(1) a genetically heterogeneous disorder(s)(2) featuring excessive skeletal mass caused by defects intrinsic to osteoclasts.(3) Engraftment of healthy hematopoietic stem cells engenders osteoclast precursors that fuse and can then remove skeletal tissue.(1,3) On the other hand, disorders caused by disturbances intrinsic to osteoblasts or chondrocytes(4,5) are challenging because these cells derive from mesenchymal stem cells.(6,7) Introduction and engraftment of osteoblast and chondrocyte precursors using marrow transplantation are technical concerns.(8–10)

Hypophosphatasia (MIM 146300, 241500, 241510)(11) is an ideal model to test marrow cell transplantation for diseases caused by defective osteoblasts and chondrocytes.(12,13) In this rare inborn error of metabolism, mutations deactivate the gene (located on chromosome 1p36.1-34) that encodes the tissue-nonspecific isoenzyme of alkaline phosphatase (TNSALP).(12,13) The biochemical hallmark is subnormal alkaline phosphatase (ALP) activity in serum (hypophosphatasemia).(13) TNSALP deficiency in osteoblasts and chondrocytes impairs skeletal mineralization, leading to rickets or osteomalacia.(12) Remarkably, only bone and dentition seem to be compromised directly.(13) Nevertheless, there is an extreme range of expressivity of hypophosphatasia, spanning from a perinatal form (often causing stillbirth from an unmineralized skeleton) to an “odonto” form (featuring only premature loss of teeth because of insufficient cementum).(13) Severely affected infants and children inherit hypophosphatasia as an autosomal recessive trait.(14,15) More than 100 different TNSALP mutations have been documented in patients worldwide.(14–19)

TNSALP is normally rich in osteoblasts and chondrocytes where it is coupled to the cell surface and functions as an ectoenzyme.(12,13) In hypophosphatasia, three phosphocompounds accumulate extracellularly: phosphoethanolamine (PEA), inorganic pyrophosphate (PPi), and pyridoxal 5/-phosphate (PLP).(13,20–22) Each compound seems, therefore, to be a natural substrate for TNSALP.(12,13,23–25) Notably, PPi is an inhibitor of hydroxyapatite crystal growth.(26,27) Hence, PPi build-up could account for the impaired skeletal mineralization.(12,28,29) In fact, the clinical severity of hypophosphatasia is reflected by the magnitude of these biochemical disturbances, radiographic abnormalities, and histopathologic changes.(13,30,31) Accordingly, a considerable variety of parameters can assess experimental treatments for this disorder.(13,20)

Currently, there is no established medical therapy for hypophosphatasia.(13) Trials of enzyme replacement using intravenous infusions of various types of ALP have failed to significantly reduce endogenous levels of PEA, PPi, or PLP, or to cause unequivocal clinical or radiographic improvement.(32–34) ALP activity must be increased, it seems, not in the circulation, but in the skeleton itself.(12,13,20) Hence, we are assessing dietary phosphate restriction as a means to achieve this goal in relatively mild cases by diminishing competitive inhibition of TNSALP by inorganic phosphate (Pi).(35)

Here, for a severely affected patient with the infantile form of the disease, we report the first test of bone marrow transplantation for hypophosphatasia.


Case report

This white girl was born to unrelated parents after an uneventful, full-term pregnancy. She weighed 3.3 kg (50th percentile) at birth, was breastfed for 1 week, and then consumed Enfamil ProSobee (Mead-Johnson Bristol-Meyer Squibb, Inc., Evansville, IN, USA).

She seemed healthy until 2.5 months of age, when she developed unexplained fever, weight loss, and failure to thrive. Investigations elsewhere revealed hypophosphatasemia (serum ALP activity, 49 IU/liter; normal for age, 143-370 IU/liter) and mild rickets.

At 4 months of age, she was admitted to Shriners Hospitals for Children, St. Louis, MO, USA. Developmental milestones seemed normal, but her weight was at the 5th percentile. She had mild frontal bossing and slight proptosis, bitemporal narrowing, a large anterior fontanel, open posterior fontanel, and patent sagittal and lambdoidal sutures. Her chest was not deformed, but the ribs were somewhat flared, although not configuring a “rachitic rosary.”

Biochemical studies showed that serum calcium and Pi levels were not low (Table 1). In fact, the concentrations of ionized calcium (Ca2+) and Pi were slightly elevated. Mild hypercalcemia likely explained the subnormal parathyroid hormone (PTH) level, enhanced serum Pi concentration, and hypercalciuria (Table 1). (These common biochemical findings in severe hypophosphatasia probably reflect impaired skeletal uptake of Ca2+.(13)) Serum ALP activity was low (93 IU/liter; normal, 133-347 IU/liter), and urinary PEA and PPi levels and plasma PLP concentration were all markedly elevated (Table 1).

Table Table 1.. Laboratory Findings at Diagnosis
original image

Radiographic skeletal survey documented worsening skeletal disease over 1.5 months. Metaphyses of the proximal humeri, wrists, knees, and ankles were more frayed (see below). Cranial bones were undermineralized, and the sutures were widely patent. However, calcification of the vertebrae and ribs, except at their ends, appeared normal without fractures. We hoped that with close monitoring she would spontaneously improve, as sometimes occurs in infantile hypophosphatasia (Fig. 1).(13,36,37)

Figure FIG. 1..

Spontaneous improvement in a case of infantile hypophosphatasia. (A) When the patient was 1 month old, this radiograph, taken for bowed legs, showed widened and poorly mineralized metaphyses, especially in the tibia (arrows). A “tongue” of radiolucency, characteristic of hypophosphatasia, projected into the diaphysis from the proximal metaphysis (black arrow). (B) At 6 years of age, without attempted therapy or interval follow-up, the growth plates were narrow and the bones were well mineralized.

By 6 months of age, however, there was considerable further clinical deterioration, including dyspnea. Radiographs up to 7 months of age documented advancing skeletal disease. Scoliosis and rib fractures narrowed her thorax (Fig. 2). Hypomineralization had progressed. She had multiple appendicular fractures (Fig. 3). Biochemical studies, however, revealed no additional explanation for worsening rickets.

Figure FIG. 2..

Chest radiographs. (A) At 2.5 months of age, the patient's ribs were well mineralized, and the thorax was not deformed. Irregularity of the metaphysis of the proximal humeri was present but not well shown. (B) At 7 months of age, just before BMT, the ribs were deformed, hypomineralized, had fractures, and were uneven and widened at their anterior ends. Scoliosis had developed. The glenoid fossae were irregular. Proximal humeral metaphyseal unevenness and widened growth plates were not well shown.

Figure FIG. 3..

Right wrist (PA radiographs). (A) At 7 months of age, just before BMT, there were distinctive metaphyseal irregularities, considerable skeletal hypomineralization, and fracture deformity of the ulna. (B) At 11 months of age, 3 months after BMT, there was remarkable improvement in the metaphyseal unevenness with dense metaphyseal calcification. (C) At 18 months of age, 10 months after BMT and just before SCB, significant skeletal deterioration had occurred with widened growth plates, metaphyses that were more uneven, loss of dense metaphyseal calcification (previous dense areas may have migrated toward the diaphysis), and residual fracture deformities. (D) At 32 months of age (11 months after SCB), the metaphyseal irregularities and skeletal mineralization had improved dramatically.

In our experience, sequential radiographic changes prognosticate most reliably for infantile hypophosphatasia.(13) Relentless and significant skeletal deterioration, without an additional reason, heralds a lethal outcome from pneumonia and pulmonary insufficiency. Accordingly, subcutaneous injections of synthetic salmon calcitonin (CT) therapy (Calcimar; Adventis Laboratories, Parsippany, NJ, USA) were begun with the hope of blocking bone demineralization,(37) but follow-up radiographs soon after showed no improvement.

At 8 months of age, we explained to the patient's parents that the clinical and radiographic deterioration seemed solely from hypophosphatasia and reviewed our previous treatment failures with enzyme (ALP) replacement.(32–34) The rationale for bone marrow transplantation (BMT), including our encouraging preliminary findings using BMT in the tnsalp knockout mouse,(38) was discussed. They elected to have their daughter undergo BMT, although the procedure was experimental for hypophosphatasia. Informed written consent was obtained according to the policies of the Institutional Review Board, Duke University Medical Center, Durham, NC, USA.

As detailed below, our patient had two types of marrow cell transplantation. At 8 months of age, she received T-cell-depleted bone marrow from her sister. At 21 months of age, a “stromal cell boost” (SCB) was performed after the sister's marrow was expanded ex vivo to enrich for stromal cells.

To establish the cause of our patient's skeletal disease, we performed mutation analysis of both of her TNSALP alleles at Washington University School of Medicine, St. Louis, MO, USA.(17)

BMT and SCB were carried out by the Pediatric Stem Cell Program at Duke University Medical Center, Durham, NC, USA.

TNSALP gene studies

Leukocyte DNA, obtained after informed written consent and before BMT, was amplified by polymerase chain reaction (PCR) and screened by denaturing gradient gel electrophoresis (DGGE) for TNSALP mutations.(16) In the absence of major deletions, DGGE can reveal defects in any of the 11 TNSALP coding exons or the adjacent splice sites of this single copy gene, and DGGE has 100% efficiency for detecting such compound TNSALP mutations in hypophosphatasia.(17)

When DGGE screening indicated two different mutated TNSALP exons (4 and 5) in the patient, each was reamplified by PCR and sequenced on both strands. The parents' DNA was then processed in the same manner.(17)

Bone marrow transplantation

The marrow donor, our patient's healthy older sister with normal serum ALP activity, was a 4/6 HLA antigen match, but genetically identical for the DRB1 locus because of a crossover event in the patient (Table 1).

Preparation for BMT consisted of 16 doses of busulfan at 40 mg/M2 (12 mg) orally every 6 h, achieving therapeutic plasma levels (days −9 to −6), 4 doses of cytoxan at 50 mg/kg/dose/day iv (days -5 to -2), and 3 doses of antithymocyte globulin at 30 mg/kg/dose/day iv (days -3 to -1). Busulfan pharmacokinetics were documented after dose 2, and dosage was adjusted after dose 6 to target a Css of 600-900 ng/ml. Trimethoprinosulfate and ganciclovir were administered intravenously for 8 days pretransplant as prophylaxis for Pneumocytis carini pneumonia and cytomegalovirus infection, respectively.

On day 0 (April 23, 1997), the sister's bone marrow was harvested using standard techniques. Mononuclear cells were isolated on a Baxter CS3000 cell separator and depleted of T-cells using T10B9 monoclonal antibodies (Medimmune Inc., Gaithersburg, MD, USA) and baby rabbit complement (C-six Diagnostics Inc., Mequon, WI, USA). The patient received 2.1 × 108 mononuclear cells/kg recipient weight, 8.4 × 106 CD34+ cells/kg, and 1.3 × 107 CD3+ cells/kg.

She was cared for under high-efficiency particulate air filtration (HEPA) filtration and positive pressure ventilation. Routine supportive care included posttransplant acyclovir, low-dose amphotericin-B, intravenous immunoglobulin (IVIG), total parenteral nutrition (TPN), low-dose heparin, and transfusions with leukocyte-depleted, irradiated, packed red blood cells, and platelets. Granulocyte-colony stimulating factor (G-CSF; Neupogen; Amgen, Inc., Thousand Oaks, CA, USA) was administered at 10 mg/kg iv from days 0 to +18. Methylprednisolone and cyclosporine were given to prevent acute graft-versus-host disease (GVHD).

As detailed below, the patient improved both clinically and radiographically until 6 months post-BMT, when deterioration began. Therefore, the SCB was performed.


Because hypophosphatasia is a disease of marrow stroma and our patient deteriorated clinically between 6 and 9 months post-BMT in association with partial loss of donor chimerism, bone marrow was again harvested from her sister and expanded ex vivo for SCB. The SCB was administered 13 months post-BMT, when the patient was 21 months of age.

To enrich for harvested osteoblast and chondrocyte progenitors, the mononuclear cells were isolated by density gradient centrifugation from 22 ml of bone marrow (546 × 106 cells). They were then expanded ex vivo in the presence of PIXy, Flt-3-ligand, and Epo for 12 days in a single-pass, stromal-based, closed-perfusion, culture chamber using the AastromReplicell System (Aastrom Biosciences Inc., Ann Arbor, MI, USA). Preclinical studies provided the methodology to maximize both hematopoietic cell growth and stromal layer development.(39)

After expansion, the cells were harvested, washed, and injected intravenously without additional myeloablation or immunosuppression therapy, although the patient continued cyclosporine prophylaxis.

Biochemical studies


Urinary PEA was measured in the Metabolic Genetics Laboratory of the Department of Pediatrics, Washington University School of Medicine, St. Louis, MO, USA, using a Beckman 7300 analyzer and reagents and procedures provided by the manufacturer (Beckman, Palo Alto, CA, USA).


Urinary PPi was quantitated using a modification of the radiometric uridine diphosphoglucose pyrophosphorylase method.(40) The established sensitivity is in the picomole range.(40)


Plasma PLP was assayed using the cation-exchange, high-performance liquid chromatography method of Mahuren and Coburn.(41)

Bone-specific ALP:

Serum levels of the bone isoform of TNSALP (BALP) were quantitated with an Alk Phos B kit (Metra Biosystems, Mountain View, CA, USA).


TNSALP mutation analysis

Our patient is a compound heterozygote for two different TNSALP missense mutations (Fig. 4). In exons 4 and 5, arginine → histidine substitutions, G212A (Arg54His) and G407A (Arg119His), were inherited from her father and mother, respectively (Fig. 5).

Figure FIG. 4..

TNSALP mutations in the patient. DNA sequencing gel showing mutations in exons 4 and 5 of the patient. For exon 4, the complementary sequence is shown. The arrows indicate position of mutations.

Figure FIG. 5..

TNSALP mutations in the parents. DNA sequencing results showing (A) the exon 4 mutation in the father and (B) the exon 5 mutation in the mother. Arrows indicate site of mutations. Sequencing chromatograms show Arg54His and Arg119His TNSALP mutations in exons 4 and 5 of the patient's father and mother, respectively, that were both transmitted to the patient.

The paternal G212A mutation (Arg54His) was excluded as a polymorphism using the allele-specific oligonucleotide (ASO) assay(17) and was not detected in 168 control alleles. This mutation has been reported once, in a lethal case of hypophosphatasia.(42) Furthermore, two different mutations that would cause different substitutions involving this same amino acid at position 54 (Arg54Cys, Arg54Pro) have occurred in lethal cases.(14) Another substitution at position 54 (Arg54Ser) resulted in the “childhood” form of hypophosphatasia.(43) Hence, the TNSALP sequence encoding amino acid 54 is a “hot spot” for change.(14) In fact, Arg54 is conserved both in the human placental ALP isoenzyme and in E. coli ALP, indicating that Arg 54 is important for TNSALP function.(14)

The maternal G407A mutation (Arg119His) has appeared in infantile and childhood hypophosphatasia.(42,44) However, Arg119 is not conserved among various species.(42)

Bone marrow transplantation

Except for mild mucositis and one temperature spike, the early post-BMT course was uncomplicated. The patient engrafted neutrophils on day +9, with last platelet and erythrocyte transfusions on days +12 and +19, respectively. She manifested neither acute nor chronic GVHD and was discharged from hospital at day +20. Clinic follow-up was unremarkable. Methylprednisolone was weaned between days +19 and +100 (Fig. 7). The salmon CT injections were changed to nasal spray CT (Miacalcin; Novartis Pharmaceuticals Corp., East Hanover, NJ, USA) for 5 months (two puffs every 5 days) until 1 year of age.(37) She wore a torso brace for scoliosis. TPN was changed to a soft diet on day +42 because her teeth had not yet erupted.

Figure FIG. 7..

Pharmacologic treatments associated with marrow cell transplantation procedures. Three drugs (synthetic salmon calcitonin, cyclosporine, and glucocorticoids) that could influence mineral and skeletal metabolism were administered to the patient near the times of BMT and SCB.

At day +100, HLA typing was consistent with full donor chimerism, that is, 100% peripheral blood cells of donor origin. Skeletal radiographs, first obtained 3 months after BMT, revealed striking improvements. Substantial healing of rickets was accompanied by generally better skeletal mineralization (Fig. 3). Metaphyseal remineralization had occurred diffusely; not only at the zone of provisional calcification. There were no new fractures.

Four months after BMT, several developmental milestones were regained. However, no biochemical corrections had occurred (Fig. 6). In serum, total ALP and BALP activity remained low. Furthermore, elevated plasma PLP concentrations persisted. Urinary PEA remained high, and perhaps most importantly, PPi levels had not decreased.

Figure FIG. 6..

Biochemical changes of hypophosphatasia. Despite clinical and radiographic improvements (transient after BMT and prolonged after SCB), there has been no correction lifelong of low ALP (or BALP) activity in serum, elevated PLP levels in plasma, or increased PEA or PPi levels in urine. Hatched areas depict normal ranges for age.

At 14 months of age, 6 months after BMT, the patient again showed arrest of developmental milestones with clinical and radiographic deterioration. Reduced donor chimerism engendered ∼10% circulating host leukocytes. Return of severe rickets included worsening chest deformity, greater scoliosis, and new rib and long bone fractures (data not shown). Routine cyclosporine and prophylactic gammaglobulin therapy continued (Fig. 7). Cyclosporine was given, however, at varying doses 9-27 months post-BMT. One year post-BMT, during this period of clinical and radiographic deterioration, she resumed salmon CT, which was administered nasally until age 22 months (just after her SCB; Fig. 7).

At 1.5 years of age, her weight was 5.7 kg and length was 62 cm. Radiographs compared with 11 months of age showed worsening skeletal disease with extreme rachitic widening of physes, especially in major long bones (Fig. 3). However, demineralized metaphyseal areas did not reflect re-emergence of rickets because there had been little longitudinal growth. Instead, skeletal hypomineralization was worse diffusely, and many new fractures affected her arms, ribs, and legs. Scoliosis was more pronounced. Further loss of donor cells was disclosed by ∼80% donor chimerism in blood and bone marrow. Possibly, reduction in donor chimerism, with loss of engraftment of donor stromal cells forming osteoblasts, explained her deterioration. We hypothesized that infusion of donor marrow stromal cells would be beneficial.


After informed written consent, donor marrow-derived stromal cells were expanded ex vivo engendering a 5-fold increase in total nucleated cells and a 73-fold increase in colony forming units (CFU-GM), partially effecting stromal cell proliferation.(39)

At 21 months of age, after premedication with steroids and diphenhydramine, the patient received (intravenously over 30 minutes) expanded marrow cells at 1.72 × 109 total nucleated cells (delivering 2.92 × 107 cells/kg; Fig. 7). Chemotherapy ablation was not given. The cells had not undergone additional proactive T-cell depletion, because T-cell expansion is suppressed by hydrocortisone in the AastromReplicell system.(39) Moderate respiratory distress with hypoxia, responsive to O2 supplementation, resolved 48 h later. There were no other clinical sequelae. Cyclosporine therapy continued to prevent marrow cell rejection and GVHD.

At 2 years of age (3 months later), tracheal intubation was necessary to manage respiratory distress from parainfluenza pneumonia. After a 1-month convalescence, her pulmonary compromise caused by scoliosis intermittently necessitated 0.5 liter/min nasal O2.

Radiographs after SCB were first obtained 6 months later. Again, remarkable skeletal improvement had occurred. At 34 months of age, growth plates were less widened, with diminished metaphyseal irregularity, and bone mineralization had generally improved (Fig. 3).

At 3 years of age, the patient was small with severe scoliosis, but was ambulating with a walker, fine-motor skills were good, and she had excellent intellectual development. Prednisone therapy, given by her pediatrician for respiratory difficulties after the pneumonia, was being tapered using alternate-day dosing. Chimerism studies showed that >90% of circulating leukocytes were donor cells.

At 3.75 years of age, further generalized skeletal improvement included continued healing of old fractures (Fig. 8). She walked freely without pain until a bowed right femur fractured during a fall. For some time afterward, she preferred to crawl. She had grown, especially during the previous 6 months off of prednisone therapy, but both weight and height remained ≪5% (9.3 kg, 83 cm). A Flovent pump was used twice a day, and an albuterol inhaler was used occasionally. She took amoxicillin each day prophylactically and used 0.5 liter O2/min; otherwise, there was some O2 desaturation from her scoliosis. Six anterior deciduous teeth had been lost prematurely—a common complication of hypophosphatasia.(13,44) Reportedly, her adult teeth were better mineralized. Radiographs showed premature closure of the sagittal, lambdoidal, and squamosal sutures—also common in hypophosphatasia.(13,45)

Figure FIG. 8..

Radiographs in early childhood. (A) At nearly 3 years of age, 1 year after SCB, the lower extremities were deformed by bowing of the major long bones, but rachitic disease was no longer pronounced and the skeleton was more uniformly mineralized. (B) At 3.75 years of age, 2 years after SCB, there was osteopenia and bowing deformity, but rachitic disease was now minimal. A small “tongue” of radiolucency projecting from the physis into the metaphysis of the distal ulna was more in keeping with the childhood form of hypophosphatasia.(13)

At 4 years of age (3 years after SCB), she was bright and talkative. Severe scoliosis (80% curve) persisted. Some significant worsening of her skeletal disease had occurred during the previous year. There were more metaphyseal irregularities and lucencies. In the lower limbs, bowing had increased. The femur fracture was, however, solidly bridged and modeled.

Now, at 6 years of age, she remains quite small (weight, 10.2 kg; height, 87 cm). No immunosuppressants are taken, but she continues 0.5 liter O2, receives amoxicillin prophylaxis and Flovent, and uses an albuterol inhaler occasionally. There are no skeletal or joint pains. She ambulates with a walker. The right leg is bowed. Her 90% scoliosis seems stable, and pulmonary status seems somewhat better. Restriction fragment length polymorphism (RFLP) studies show >95% donor cells. Radiographs disclose improved mineralization, better-modeled long bones with more defined cortices, and regular-appearing metaphyses compared with 1.33 years before. However, there is closure of all cranial sutures.

Despite the clinical and radiographic improvement after BMT and SCB, biochemical disturbances reflecting hypophosphatasia remain essentially unchanged (Fig. 6). Furthermore, circulating concentrations of calcium and Pi remain normal (Fig. 9).

Figure FIG. 9..

Lifelong serum calcium and Pi levels. Skeletal changes cannot be explained by fluctuations in circulating calcium (▾) or phosphate (•) levels (hatched areas are normal ranges for age: top, calcium; bottom, phosphate).


Hypophosphatasia is an excellent model for evaluating marrow cell transplantation for skeletal disease caused by defects intrinsic to non-hematopoietic, marrow-derived cells, that is, osteoblasts or chondrocytes.(12,20) Assessments of clinical, radiographic, and histopathologic manifestations, as well as quantitation of TNSALP activity and substrate levels endogenously, can all help to interpret the outcome.(12,20)

In our patient, compound heterozygosity for two different TNSALP missense mutations confirms that deficient ALP activity caused her skeletal disease. Indeed, the aberrations (albeit in novel combination) are each associated with severe hypophosphatasia.(14,42) Nevertheless, accumulation of PPi, caused by impaired TNSALP-mediated PPi hydrolysis at the surface of osteoblasts and chondrocytes or their matrix vesicles, is the likely proximate explanation for her defective skeletal mineralization.(12,13,46) PPi inhibits hydroxyapatite crystal growth.(12,13,20,26) Hence, reduction of endogenous PPi levels seems to be a key therapeutic objective for hypophosphatasia.(26,46)

Unfortunately, neither Mg2+ cofactor(47) nor Zn2+(48) supplementation administered orally to enhance TNSALP activity or injections of a PTH fragment(49) to stimulate TNSALP biosynthesis, have been of benefit to hypophosphatasia patients. Furthermore, we failed to significantly improve infantile hypophosphatasia by intravenously infusing various types of ALP.(32–34) This approach to enzyme replacement therapy can correct hypophosphatasemia, but does not overcome TNSALP substrate accumulation. Hence, circulating ALP seems irrelevant to the skeleton.(32–34) Greater ALP activity in osteoblasts, chondrocytes, etc. at calcifying surfaces will be necessary to restore bone growth and mineralization. In fact, there is evidence that ALP must be tethered to the external side of the plasma membrane as either a homodimer or homotetramer to function.(12,20)

BMT followed by SCB seems to explain the survival and then long-term advances, respectively, of our patient. Relentless skeletal deterioration, a harbinger of a lethal outcome for hypophosphatasia, occurred during the 6 months after diagnosis until BMT. Soon after BMT, significant clinical and radiographic gains occurred. Bone densitometry or histomorphometry were not necessary to document the progress, because the X-ray changes were so striking. In fact, this time frame after BMT typically features bone loss caused by short-term suppression of skeletal formation and acceleration of bone resorption.(50) Nevertheless, our patient's gains after BMT were fleeting. Accompanying partial return of host hematopoiesis, skeletal deterioration recurred 6 months afterwards. Fortunately, rapid, substantial, and prolonged improvement followed SCB.

Our cumulative findings seem puzzling, however, because there were no biochemical corrections. Low ALP and BALP activities in serum, elevated PLP concentrations in plasma, and high PEA or PPi levels in urine have persisted. Unfortunately, our patient's clinical status has seemed too precarious to biopsy her skeleton. We do not know what percentage of her stromal cells, osteoblasts, chondrocytes, etc., might now be donor-derived. Although a combination of factors could account for our findings, six potential explanations are assessed below.

Spontaneous changes:

In our experience with infantile hypophosphatasia, that is, skeletal disease recognized after birth but before 6 months of age,(13) the patient's course cannot be predicted at initial examination (Fig. 1). Prognostication is based on the sequential clinical findings, and especially the radiographic changes.(13,45) Reportedly, 50% of these patients survive and 50% die.(51) Early publications indicate that substantial gains are more likely in those who live past infancy.(51) Nevertheless, striking fluctuations in disease severity typically do not occur.(49,52) However, in one exceptional patient, transient marked improvement of all features of hypophosphatasia followed intravenous infusions of pooled blood-bank plasma,(49) somehow overcoming homozygosity for missense mutations in TNSALP.(53) Unfortunately, the observations for this patient remain an enigma.(49,53) Also, why the skeleton can appear relatively normal at birth in patients with the infantile form of hypophosphatasia but then deteriorate is unexplained.(13) Protection in utero is probably not occurring because perinatal hypophosphatasia can cause almost complete absence of fetal calcification.(13)

Our patient's skeletal disease has fluctuated considerably. Unexplained alterations in the radiographic findings have occurred during the past 2 years; however, the previous improvements closely coincided with the marrow cell infusions. Rickets and generalized hypomineralization corrected transiently after BMT, recurred when host hematopoiesis partly resumed, but then improved after SCB. Hence, spontaneous amelioration seems unlikely for our patient.(3,13,36)

Cyclosporine therapy:

Cyclosporine treatment can accelerate skeletal remodeling,(54,55) and it began at BMT for our patient. Possibly, this immunosuppressant stimulated osteoblasts and chondrocytes, thereby overcoming the defective mineralization.

However, the biochemical findings (i.e., no correction in serum ALP activity) fail to support this hypothesis. Furthermore, cyclosporine therapy was continued during the post-BMT deterioration and until after the post-SCB improvement (Fig. 7). In fact, cyclosporine can adversely affect the skeleton, contributing to the osteoporosis that follows solid organ transplantation.(54,55) Accordingly, we doubt that cyclosporine effects explain our observations.

Glucocorticoid administration:

Skeletal improvements in our patient seem to coincide with glucocorticoid administration. Methylprednisolone and prednisone were given after BMT to prevent GVHD, and prednisone was administered for 1 year beginning a few months after SCB for respiratory difficulties exacerbated by viral pneumonia (Fig. 7). In fact, several early, brief case reports suggest that glucocorticoids are helpful for hypophosphatasia,(32,36,56) although no explanation is offered.(13)

Glucocorticoids can mitigate the hypercalcemia that often complicates infantile hypophosphatasia, likely by blocking gastrointestinal absorption of calcium. Although dexamethasone promotes differentiation of precursor cells to osteoblasts in vitro,(57,58) glucocorticoids are detrimental to osseous tissue.(55) Suppression of bone formation reflects the dose of prednisone used after BMT for other conditions.(50) Indeed, our patient's skeletal deterioration after BMT occurred after maximum prednisone dosing. Furthermore, with resumption of glucocorticoid therapy after her pneumonia, dosing was tapered and then discontinued at about 3.5 years of age when skeletal radiographic improvement was still occurring. Some skeletal decline occurred by 4 years of age, during the previous year off prednisone therapy, but had corrected by age 6 years. Accordingly, there is some, but not compelling, temporal association between glucocorticoid use and our patient's clinical and radiographic advances. In fact, glucocorticoid-associated benefit to the skeleton is not a consistently reported finding for hypophosphatasia(51) or something that we have encountered.

Salmon calcitonin therapy:

We initiated salmon CT injections for our patient on learning of the case report of Barcia et al.(37) using this hormone to control hypercalcemia and to block skeletal resorption in a patient with infantile hypophosphatasia. For our patient, CT therapy accompanied the early post-BMT improvements, ceased about 1 month before the post-BMT deterioration, and resumed before the post-SCB improvements (Fig. 7).

However, radiographic improvements did not occur in our patient during the several weeks just before BMT. Furthermore, CT treatment ended soon after SCB, yet there was prolonged benefit. In fact, our personal experience and case reports by other investigators suggest little skeletal gain using CT for infantile hypophosphatasia.(59) Therefore, CT therapy seems an unlikely explanation for our observations.

Increased ALP in leukocytes:

Leukocyte ALP (LALP) can be deficient in hypophosphatasia,(13) suggesting that at least some is an isoform of TNSALP.(60,61) Consequently, engraftment of healthy donor marrow hematopoietic cells should augment TNSALP levels in the hypophosphatasia skeleton.

However, LALP is sequestered within neutrophil granules, apparently for release during phagocytosis, etc.(61) This intracellular distribution contrasts with cell-surface TNSALP in osteoblasts and chondrocytes.(62) LALP is probably inaccessible to PEA, PPi, and PLP in extracellular fluid because these phosphorylated compounds cannot cross plasma membranes.(62) In fact, BMT in our patient led to rapid, significant, and essentially permanent engraftment of donor hematopoietic cells, yet skeletal improvement fluctuated. However, her LALP activity, initially normal after BMT when first measured, inexplicably waned (normal range, 15-100%) to subnormal values, that is, 33 at 14 months of age, 10 at 1.5 years of age, 4 at 2.83 and 4.58 years of age, and 2 at 6 years of age, despite substantial, prolonged benefit after SCB. Accordingly, it is unlikely that LALP explains our patient's advances.

Engraftment of mesenchymal stem cells augmenting TNSALP activity in osteoblasts and chondrocytes:

We favor engraftment of donor marrow stromal cells to explain our findings. Beneficial skeletal effects could occur because TNSALP is properly targeted. In this scenario, serum ALP activity might not be corrected because the liver contributes importantly to circulating TNSALP and would not be altered by BMT or SCB. Substrate accumulation in blood and urine also might not diminish.

In fact, tnsalp null (−/−) mice, developed in two independent laboratories,(63,64) recapitulate infantile hypophosphatasia.(65) In 1996, our preliminary findings indicated that BMT improved their skeletal disease without changing circulating ALP activity.(38) Notably, a few ALP-replete osteoblasts appeared.(38) Subsequently, in 1998, transfection studies of several hypophosphatasia TNSALP missense mutations showed that ALP activity >4.1% wild-type distinguishes nonlethal from lethal cases of hypophosphatasia.(17) Although research or clinical laboratory assays of ALP activity typically are not physiologic measurements because they use artificial substrates and markedly alkaline pH,(35,66) the observations nevertheless suggested that small endogenous increments in TNSALP activity could have important clinical effects. Similarly, that same year, a murine model for osteogenesis imperfecta (OI, MIM 166210), the oim mouse, provided findings which supported this possibility.(67,68) OI is caused by dominant mutation in either of the two different genes that encode the type I collagen heterotrimer. Hence, OI too is a disease of osteoblasts, but it involves quantitative and often qualitative defects in type I collagen, leading to “brittle bones.”(11)Oim bone mass increased after BMT despite engraftment of few donor osteoblasts.(67,68) Additionally, in 1999, three children with OI reportedly benefitted from allogeneic BMT.(69) Their growth, skeletal density, bone histology, and fracture frequency were said to improve during 3-6 months of follow-up despite engraftment of only 1.5-2.0% donor mesenchymal cells (osteoblasts).(69) We note that parameters for assessing OI are fewer compared with hypophosphatasia,(70) and BMT for OI remains controversial.(70–74)

Our patient's radiographic studies showed that not only trabecular bone mineralization improved, but of interest, so did cortical bone thickness and endochondral ossification. Increased ALP activity in medullary osteoblasts would help mineralize trabecular bone. However, cortical bone formation, mediated by subperiosteal osteoblasts or growth plate improvements dependent on chondrocytes, were especially remarkable findings. Perhaps, healing of our patient's rickets reflected some uncharacterized systemic effect of the marrow cell transplantations. Alternatively, it may be that TNSALP replete subperiosteal osteoblasts and chondrocytes emanated from donor stromal cells. In fact, chondrocyte precursors are derived from mesenchyme and marrow stroma is highly regenerative when transplanted in vivo into diffusion chambers.(6) CFU-F give rise to osteoblasts, myoblasts, adipocytes, and chondrocytes.(6)

The especially favorable and prolonged response after SCB compared with BMT seems attributable to delivery of relatively large numbers of donor stromal cells. In fact, there have been attempts in humans to transplant stromal cells after ex vivo proliferation in the autologous setting.(39) The AastromReplicell System expands bone marrow for augmentation of autologous transplantation and was used for our patient's SCB. Ex vivo expansion of bone marrow in small, clinical-scale-regulation perfusion experiments using this system can significantly increase numbers of total nucleated cells, CFU-GM, and long-term culture-initiating cells.(39) In a clinical study using this system, CFU-F (stromal progenitors) increased, on average, 40-fold.(75) Furthermore, recently, infusion of mesenchymal stem cells expanded ex vivo has benefitted metachromatic leukodystrophy and Hurler syndrome, without evidence of significant engraftment.(76)

Mixed chimerism may be adequate to cure a number of conditions, including inborn errors of metabolism or other genetic defects, in which non-myeloablative therapy combined with immunosuppression establishes donor cells.(76) However, mixed chimerism seemed detrimental to our patient, perhaps reflecting loss of marginal numbers of stromal cells. We do not know from TNSALP gene studies whether the donor sister, whose serum ALP activity is normal, is a hypophosphatasia carrier. If so, perhaps the response to marrow cell transplantation would have been better.

In summary, BMT followed by improved skeletal mineralization apparently rescued an 8-month-old girl with infantile hypophosphatasia. However, worsening skeletal disease recurred after partial return of host hematopoiesis. Prolonged clinical benefit and bone and cartilage remineralization occurred after SCB. Nevertheless, there was no correction of the biochemical disturbances of hypophosphatasia. Our findings could reflect engraftment of small, but sufficient, numbers of donor mesenchymal stem cells producing ALP-replete osteoblasts and chondrocytes. Experience with this first patient to undergo marrow cell transplantation for a defect intrinsic to osteoblasts and chondrocytes is promising. Studies of additional appropriate patients with severe hypophosphatasia and the tnsalp knockout mouse will clarify if and how this procedure ameliorates this disorder.


We are grateful to Joanne Reid, MD, for referring the patient and for forwarding follow-up information. The dedicated and skilled nursing, dietary, and technical staff of the Center for Metabolic Bone Disease and Molecular Research, Shriners Hospitals for Children, St. Louis, MO made this study possible. Jonathan Jones and Patrick Finnegan performed the TNSALP mutation analyses. Becky Whitener, CPS, provided expert secretarial help. This study was supported in part by Grants 8580 and 8540 from Shriners Hospitals for Children, The Clark And Mildred Cox Inherited Metabolic Bone Disease Research Fund, and The Hypophosphatasia Research Fund.