MIA is a potential biomarker for tumour load in neurofibromatosis type 1

The mice were continuously back-crossed to wild-type C57BL/6J to minimize the variation of genetic background. The female Nf1flox mice were crossed to male Nf1flox heterozygous Prx1-Cre-positive males and the offspring genotyped as previously described [9]. Embryos and postnatal tissue samples were fixed overnight at 4°C in 4% paraformaldehyde, dehydrated through an ethanol/xylol series, and embedded in paraffin blocks. Six-micrometer sections were cut and processed for haematoxylin and eosin/Alcian blue staining and in situ hybridization.

The study was conducted with a cohort of 42 NF1 patients and 22 healthy individuals. The diagnosis of NF1 was made using National Institutes of Health criteria. The study protocol was approved by the local institutional review board, and all patients gave their informed consent. Cutaneous and subcutaneous tumours were counted or estimated in case the number was larger than 100. Plexiform neurofibromas, including internal ones, were detected by means of whole-body MRI in 30 of the 42 patients. Because of the limited resolution of whole-body MRI, lesions smaller than 3 cm in the longest diameter, which is often the case for spinal tumours, were not included. Tumour sizes were calculated using a semiautomated volumetric method, and the total internal tumour load was obtained subsequently, including PNs (Plexiform Neurofibromas), spinal tumours and internal nodule neurofibromas, but excluding cutaneous and subcutaneous tumours [19]. An age effect for cutaneous, subcutaneous and internal tumours was examined using a nonparametric Spearman's rank-correlation test.

All serum samples were prepared using a standardized protocol in the laboratory of the Department of Maxillofacial Surgery at the University Medical Center Hamburg-Eppendorf. Whole blood of each patient was kept at room temperature for 30 minutes before being spun down at 4,500 rpm for 10 minutes using a benchtop centrifuge. The supernatant was stored at -80°C in 100-μL aliquots.

In situ hybridization was performed on paraffin sections according to standard protocol [9]. Images were collected using a DMR HC microscope (Leica, Wetzlar, Germany) equipped with an AxioCam HRc camera (Zeiss, Jena, Germany) and evaluated using AxioVision 4.1 software (Zeiss, Jena, Germany).

Sections of six cutaneous and three plexiform neurofibromas, as well as seven MPNSTs, from a total of sixteen NF1 patients were stained with monoclonal anti-human MIA antibody (R&D Systems, McKinley Place NE, Minneapolis) diluted at 1:40. Sections were boiled in citrate buffer (pH 6.1) for antigen retrieval. The streptavidin-biotin method was performed using an automated staining system TechMate (Dako, Hamburg, Germany) with an implemented counterstaining. Negative controls were carried out with normal serum without the primary antibody or with antibody preincubated in access (25 ng/μl) of recombinant human MIA (Peprotech GmbH, Hamburg, Germany). Stained sections were analyzed using the BX51 microscope (Olimpus, Hamburg, Germany) and analySIS 5.0 software (Soft imaging system GmbH, Münster, Germany).

RNA was isolated from the knee cartilage of two wild-type and two Nf1Prx1 mice using peqGOLD TriFast (PeqLab Biotechnologie GmbH, Erlangen, Germany) according to the supplied protocol. cDNA was synthesized from 1 μg of total RNA with MuLV Reverse Transcriptase (Applied Biosystems, Carlsbad, CA, USA). TaqMan Universal PCR Master Mix was then performed on an ABI PRISM 7900 Cycler (Applied Biosystems) using the SYBR Green method (Invitrogen, Darmstad, Germany) according to the manufacturer's instructions. The expression level of Mia was determined in Nf1Prx1 and wild-type tissues and was equilibrated against expression of glyceraldehyde 3-phosphate dehydrogenase (GAPDH). The following primers were used: mGAPDH: 5' GGGAAGCCCATCACCATCTT 3', 5' CGGCCTCACCCCATTTG 3'; mMIA: 5' GGAGGACCTGACTCTGAAACC 3'; 5' ACTGCAGGGATAGCGGTAG 3'.

The MIA ELISA kit was purchased from (Roche Diagnostics, Indianapolis, IN, USA) and the measurements were conducted in duplicate according to the supplied protocol. Internal negative and positive quality controls were provided in each enzyme-linked immunosorbent assay (ELISA) kit and were run in triplicate in each assay.

In situ hybridization revealed expression of Mia in the cartilage of the E14.5 to E15.5 mouse embryo (Figure 1). Expression domains of Sox9, Col2a and Mia overlapped and, in the E14.5 embryo sections, demarcated cartilage anlagen of the future bones (Figure 1A). The expression of Mia was found to be more intensive in the NF1-deficient cartilage of the Nf1Prx1 mice (Figure 1B). Similar results were obtained with mouse embryos bearing cartilage-specific inactivation of Nf1 (data not shown). We next quantified Mia expression by performing quantitative real-time polymerase chain reaction (qRT-PCR). Absolute quantification was conducted on the RNA isolated from knee cartilage of two mutant and two control mice at P4. Mia transcript levels in Nf1-deficient tissue were compared to the wild-type tissue and normalized to GAPDH expression. qRT-PCR revealed a more than twofold increase of Mia expression in Nf1Prx1-deficient cartilage.

MIA was immunohistochemically detected on the paraffin sections of six cutaneous and three plexiform neurofibromas and in seven MPNSTs from NF1 patients. The typical pattern of MIA staining was a mixture of positive and negative nuclei side-by-side (Figure 2). The proportion of MIA-positive cells varied between 50% and 90%. The most intense staining was obtained in MPNSTs, which, however, represents the high density of nuclei in this type of tumour. No morphological difference was observed between MIA-positive and MIA-negative cells. On the basis of the degenerative nuclear atypia of Schwann cells, we deduced that MIA was both positive and negative in Schwann cell nuclei. MIA-positive cells were more often seen in areas of spindle-shaped cells arranged in fascicles.

MIA serum level was determined in the 42 NF1 patients and in 22 healthy individuals. The patients' ages ranged from 14 to 72 years (mean age, 36 years). The control group's ages ranged between 19 and 67 years (mean age, 40 years). An age effect was seen in the NF1 patients for the number of cutaneous tumours (P = 0.023), but not for subcutaneous tumours (P = 0.842) or internal tumours (P = 0.449). Additionally, linear regression analysis revealed an association between total internal tumour load and the number of subcutaneous tumours (P value 8.19E-17 for the F test), but not between internal tumour load and the number of cutaneous tumours.

MIA serum concentration was independent of age and sex (data not shown), but was significantly higher in NF1 patients than in healthy controls: 15.16 ± 1.26 pg/mL versus 4.54 ± 0.40 pg/mL (P < 0.001, unpaired t-test with Welch's correction) (Figure 3A). Among the 42 NF1 patients, the 27 patients with pNFs had significantly higher MIA serum concentration than the 15 patients without those tumours (P = 0.032) (Figure 3B). However, no significant difference in MIA serum level was found between the 7 and 35 patients with and without MPNSTs, respectively (Figure 3B). MIA serum level was also significantly higher in the nine and seven patients with > 100 subcutaneous neurofibromas and > 1,000 cutaneous neurofibromas, respectively, than in those without such tumours (Figures 3C and 3D). Internal tumour load was determined for 30 of the 42 NF1 patients on the basis of whole-body MRI. The patients were divided into four groups: very low internal tumour loads (0 to 100 mL; n = 16), low internal tumour loads (< 350 mL; n = 5), moderate internal tumour loads (< 1,000 mL; n = 5) and high internal tumour loads (> 1,000 mL; n = 4) (Figure 3E, left). One-way analysis of variance with the Bonferroni multiple comparison test revealed significant differences between MIA serum levels in patients with very low internal tumour loads and groups with high and very high internal tumour loads (**P < 0.01, ***P < 0.001). Also, linear regression analysis revealed an association between the total internal tumour load and MIA serum level (P value of 1.95E-7 for the F-test). The line that best predicts MIA level from values of logarithm of internal tumor load volume was identified by regression analysis: R² = ~0.64 (Figure 3E, right). These data indicate that elevated MIA serum level may be indicative of an increased internal tumour burden. Since we observed an association between total internal tumour load and the number of subcutaneous tumours, a study involving a larger cohort size is necessary to reveal the relative contributions of internal, subcutaneous and possibly also cutaneous tumours to elevated MIA levels.