1. Introduction

The World Health Organization (WHO) has declared osteoporosis to be a major public health problem [1]. This disease affects hundreds of millions of people [2] and the situation is worsening with the increase in life expectancy which favors the appearance of this pathology. Studies show that approximately one in four women and one in eight men suffer from this disease. For women over 80 years old, these proportions increase to one in two [3,4].

These worrying figures lead to the search for easier and faster techniques to perform tests in the form of screening. The work presented here is in line with this perspective by proposing UV-visible fluorescence of hair as a technique to perform preliminary measurements to X-ray absorptiometry. It should be remembered that this latter allows the clinical diagnosis of osteoporosis by providing a value of Bone Mineral Density (BMD) which is identified by a classification defined by the T-score value [5,6]. This technique, which is the most widely used nowadays, remains cumbersome and very expensive, making it unaffordable in many parts of the world, thus supporting the interest in finding new and much more accessible testing methods.

Recent measurements [7] performed on human hair using the Front Face Fluorescence Spectroscopy (FFFS) technique have shown the ability of this technique to detect the disease even at an early stage of development. The results revealed a strong correlation between a fluorescence peak centered on 485 nm and the disease. The molecules proposed to explain the behavior of this emission peak were keratin, elastin as well as vitamin D and A which fluoresce in this region. This emission peak was then proposed as a visual marker for monitoring the disease.

In order to better understand the connection between BMD and the fluorescence spectrum, measurements were made using the LIBS technique to track the behavior of mineral salt concentrations present in hair. These measurements were performed on the same hair samples used in [7]. We focused on the concentrations of calcium given its fundamental role in the regeneration of bone tissue and in the composition of hair. Indeed, the abundance of calcium in bone constitutes by definition the probe for the detection of osteoporosis through the measurement of BMD [8]. Also, its presence in hair has been documented in numerous works [9,10–14] involving the calcium-containing protein S100A3. The abundance of this protein in the cuticle should have an impact on the observed fluorescence spectra [15] and therefore should be added to the other molecules as a potential marker of osteoporosis.

2. Methodology

2.1 Sample preparation

The samples used in this work are those used in the study by Cherni et al.; [7]. In this study, hair strands of 2 to 5 cm in length belonging to 90 patients were collected near the root. Among the 90 samples, 30 belonged to patients with osteoporosis, 30 had osteopenia and 30 were healthy. The strands had been washed according to the protocol proposed by the IAEA (International Atomic Energy Agency) Advisory Group following the sequence three times water-acetone to remove dirt and grease [16]. Vanadium-free stainless-steel scissors had been used in order not to taint the samples with unwanted components. The fluorescence analysis of these strands was done by a direct measurement of the fluorescence emission following their excitation by an LED whose emission peak is located at 365 nm.

For LIBS analysis, 0.01 gram of each hair strand was cut into very small pieces and mixed with potassium bromide (KBr). The mixture was ground, homogenized and pellets of 12 mm diameter and 5 mm thickness were made with a manual RETSCH pelletizer under a pressure of twenty tons.

2.2 LIBS measurements

The details of the experimental setup used for these measurements corresponds to the one described in [17]. In short, the focusing of a laser (Nd: YAG) pulsed at 5 ns and emitting at 532 nm produces an irradiance of 52 GW/mm² high enough to generate a luminous plasma on the surface of the pellets. Part of the light from this plasma is collected and focused on the entrance slit of a high-resolution monochromator (THR 1500; JOBIN YVON, Ltd.) (0.5 Å resolution). A fast intensified charge coupled device (ICCD) (DH520–25F-03; Andor Technology, Ltd.) was used for photon detection. Synchronization between the Nd:YAG laser and the ICCD detector was ensured by microcomputer via a pulse delay generator (DG 535; Stanford Research Systems, Inc.)

For each sample, and to minimize the matrix effects, we recorded five (5) spectra by focusing the laser on different points of the pellet surface. Moreover, each of these five spectra is the result of the accumulation of twenty-five successive pulses improving the S/N ratio. The spectra were recorded 2000ns after the laser pulse with a gate width of 500 ns. Several elements (Ca, Na, Zn) with some of their lines were recorded in this study. However, we have limited our analysis to calcium because of the major role of this element in this study. We focused more precisely on the line (3933.6 Å) of the calcium ion Ca⁺ given its systematic reproducibility and the excellent S/N ratio obtained for this line.

The presence of calcium in the hair is easily identified in the spectrum thanks to the two emission lines of the Ca⁺ ion located at 3933.6 and 3968.5 Å which are systematically present in all the spectra with an excellent S/N ratio. Moreover, these two lines do not suffer from the phenomenon of self-absorption which allows to directly relate their intensity to the concentrations present in the plasma. As an example, we show on Fig. 1 the behavior of these two lines for three particular samples, one belonging to a patient with osteoporosis (green lines), another belonging to a patient with osteopenia (blue lines) and the third corresponding to a healthy patient (red lines). The values were normalized to the value of the maximum in this spectrum.

 

Fig. 1. Calcium line intensities for the three population categories: healthy, osteopenic and osteoporotic. Normalization was performed with respect to the maximum of each spectrum.

Download Full Size | PDF

3. Results and discussions

3.1 Results

For the study of the correlations with the fluorescence measurements, we limited our analysis to the most intense peak located at 3933.6 Å since the ratio between the two lines of the same sample remains constant. The maximum of each intensity peak was recorded for each of the 90 samples and the largest value of these maxima was used for the normalization of all peaks. We show in Table 1 the average “I(LIBS)” values obtained as a function of the disease level. We have also indicated the mean values corresponding to the measurements obtained by fluorescence (P4 [7]) as well as the mean values corresponding to the BMD values (T-Score) to show the correlations between these different measurements.

Table 1. Mean values of LIBS (this work), Fluorescence (P4) and T-score measurements [Habib Thameur Hospital] for the three categories of patients. RSD are given in the parentheses next to the values.

View Table

The average values already show that the intensity of the peak obtained by LIBS decreases with the health status of the patients. In other words, the hair calcium concentrations of patients with osteoporosis are on average lower than those of patients with osteopenia, which are themselves lower than those of healthy patients. This behavior is corroborated by that of BMD (T-score). It is also observed that this behavior is like to that obtained with the fluorescence values (P4).

A Principal Component Analysis (PCA) performed on the maximum intensity peaks extracted from the spectrum of the calcium line (3933.6 Å) and the sodium line (5891.6 Å) for each sample confirms the trend noted previously. The distribution of the samples in the plane defined by the first two components Dim1 and Dim2 (Fig. 2), which represent 100% of the variance, clearly shows the discrimination obtained between the three categories of patients. The contribution to Dim1 which discriminate most is equally shared between these two elements (calcium and sodium). An ANOVA test is applied and lead to a p-value = 0.00 confirming the significance of our results. It is also noted that the relative arrangement of the clusters reflects the gradual evolution of the level of disease. Indeed, the samples of patients with osteopenia are located between the group of healthy patients and those with osteoporosis.

 

Fig. 2. LIBS emission peaks for each sample in the space defined by the first two components (Dim1 and Dim2) representing 100% of the variance.

Download Full Size | PDF

To establish the correlation between the measurements of calcium concentrations obtained by LIBS and the P4 peak extracted from the fluorescence spectra, we applied the Spearman statistic to the set of values obtained for these two parameters. Figure 3 clearly shows a strong correlation between these two quantities with a correlation factor R = 0.86 and a p-value = 6.5E-7. We have also represented in Fig. 4 the correlation between BMD measurements and the calcium peak with a correlation factor R = 0.78 and p-value = 1.7E-6.

 

Fig. 3. Diagram showing the strong correlation between the P4 fluorescence peak and the peak intensity of the Ca⁺ (3933.6 Å) line.

Download Full Size | PDF

 

Fig. 4. Diagram showing the strong correlation between the Ca⁺ (3933.6 Å) peak and clinical spinal BMD measurements. The shaded area represents a 95% confidence interval.

Download Full Size | PDF

3.2 Discussion

The reduction of calcium concentration with the level of disease, revealed by the decrease of the peak intensity of the Ca⁺ (3933.6 Å) line, means therefore the reduction of the concentration of the protein S100A3 (cysteine-rich calcium binding protein) present in abundance in the hair cuticle [9,10]. To answer the question raised by the behavior of hair fluorescence and to explain the strong correlation of the P4 peak [7] with the level of disease, it is necessary to know the role of each of the elements involving: keratin, elastin, vitamin D, vitamin A and S100A3. Concerning the S100A3 protein, there is not much data on its fluorescence for an excitation λ_exc = 365 nm. Iris et al. [15] have shown variations in the fluorescence of the S100A protein as a function of the concentration of the Ca²⁺ cation for an excitation λ_exc= 385 nm leading to an emission varying from 484 to 504 nm. This emission area corresponds precisely to the position of the P4 peak in the fluorescence spectrum in [7]. This observation suggests that this protein may also play an important role in explaining the variations of the fluorescence spectrum in the vicinity of this peak. However, given the limited data available in the literature on its fluorescence (concentrations/quantum yields), it is difficult to say whether the observed changes are a direct consequence of the variation in its concentration or an indirect manifestation of it.

Also, since we do not precisely know the contributions of the suggested elements to explain the fluorescence spectrum, it is difficult to decide which will have the strongest impact on the observed spectra. The answer to the question is made even more difficult knowing the interdependence of these different elements. For example, calcium plays an essential role in the keratinization process [18] and vitamin D is the most important regulator of calcium metabolism [19,20].

A comparative study of the fluorescence of each of these elements taken separately and for the same excitation would allow to partially answering the question to isolate and propose the suitable histological marker explaining at best the variations of the P4 peak. In the meantime, this peak can still be proposed as a visual marker of the disease.

4. Conclusion

The detection of calcium in hair by LIBS technique and the evidence of the reduction of its concentration, revealed by the decrease of the intensity line Ca⁺ (3933.6 Å), according to the evolution of the disease brings a new light to explain the behaviors observed on the fluorescence spectra obtained with the FFF technique. These new measurements show that S100A3 protein can be added as a new candidate having an impact on the P4 emission peak located near λ = 485 nm [7] with a strong correlation with the disease. Pending fluorescence measurements on reference samples that would allow to trace back to histological markers, the P4 emission peak can be used as a visual marker for the detection of osteoporosis. The results presented in this work reinforce the idea that the FFF fluorescence technique seems to offer a great opportunity for the detection of osteoporosis in screening mode.