- Email: [email protected]
Accuracy of the three-window sampling method in morphometric analysis of human sural nerve
Journal of Neuroscience Methods 157 (2006) 154–157
Accuracy of the three-window sampling method in morphometric analysis of human sural nerve Vilai Chentanez, Phinidda Cha-oumphol, Athitaya Kaewsema, Sithiporn Agthong ∗ , Thanasil Huanmanop Peripheral Nerve Research Unit, Department of Anatomy, Faculty of Medicine, Chulalongkorn University, Rama IV Road, Pathumwan, Bangkok 10330, Thailand Received 12 January 2006; received in revised form 4 April 2006; accepted 6 April 2006
Abstract Morphometry has an important role in the assessment of sural nerve biopsies as a part of early detection of structural abnormalities in peripheral nerve. Various sampling methods have been used to reduce time and effort needed in the analysis of total nerve fibers but their accuracy remains controversial. We examined the accuracy of three-window sampling method in the morphometric evaluation of human sural nerve biopsies by comparing with the total fiber quantification. Three windows (0.012 mm2 each) were placed in every possible fascicle in the sections and data from all windows were pooled and analyzed for the number of myelinated axons, myelinated fiber diameter, axon diameter, myelin thickness, g ratio as well as myelinated fiber density. Means and ranges of the data from the two techniques were similar and the agreement was further confirmed by intraclass correlation analysis. These findings indicate that the three-window sampling method can be used to evaluate human sural nerve with accuracy. © 2006 Elsevier B.V. All rights reserved. Keywords: Sampling; Sural nerve morphometry
1. Introduction Morphometric analysis of the sural nerve is commonly used to provide more objective and quantitative data compared to conventional visual examination. This is crucial for more accurate and sensitive diagnosis of peripheral nerve diseases. Since it is an energy- and time-consuming procedure to evaluate all the myelinated nerve fibers in the whole section despite an application of computer-assisted image analysis software, various sampling methods have been developed and used (Dyck et al., 1980; Jacobs and Love, 1985; Behse, 1990; Cai et al., 2002; Lindemuth et al., 2002). Systematic sampling is more preferable compared to random and fascicle sampling due to the non-uniform distribution of nerve fibers shown in previous studies (Dyck et al., 1984, 1986; Saxod et al., 1985). However, the accuracy of the sampling techniques to provide valid representative data of the whole myelinated fiber population is still controversial. Tang and Ebbesson (1972) found that approxi-
∗
Corresponding author. Tel.: +66 2 2564281; fax: +66 2 2527028. E-mail address: [email protected] (S. Agthong).
0165-0270/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jneumeth.2006.04.001
mately 50% of the total fascicular area was sufficient to obtain accurate data in cervical sympathetic nerve trunks. Saxod et al. (1985) suggested that the density of myelinated fibers in human superficial peroneal nerve should be calculated by counting all fibers. Cai et al. (2002) proposed that it was necessary to quantify all the myelinated fibers in order to obtain reliable morphometric data of the sural nerve. Despite the above evidence recommending the use of total count, laborious work associated with this technique renders the accurate sampling method still necessary for assessment of the sural nerve. One of the sampling techniques, the three-window sampling, has been used to evaluate nerve regeneration in sciatic nerves of rats and primates (Sterne et al., 1997; Ahmed et al., 1999). The method uses three randomly chosen fields under ×40 objective. However, this sampling technique has not been employed to evaluate human nerves and its accuracy has not been tested. Hence, in this study, we examined the accuracy of the threewindow sampling method compared with the count of total myelinated fibers in human sural nerve in various morphometric parameters and found that the results from both techniques were comparable. As a result, this method can be used to obtain representative morphometric data of human sural nerve.
V. Chentanez et al. / Journal of Neuroscience Methods 157 (2006) 154–157
155
2. Materials and methods 2.1. Subjects Sural nerves were obtained from 78 subjects within 24 h following death. These cases were included with approval from the Department of Forensic Medicine, Faculty of Medicine, Chulalongkorn University. There were 66 males and 12 females and their age range was 20–60 years. This range of age was used due to the report of the mature morphometric details of the sural nerve (Jacobs and Love, 1985). The included subjects must have no known causes or conditions of peripheral neuropathy. Delay duration until removal of nerve after death was 1 h at minimum and 22 h at maximum. The average age of all 78 subjects was 37 years. The most common cause of death was an accident, found in 33 cases (27 males and 6 females). The mean duration after death until the removal of the nerve was approximately 10 h. 2.2. Sural nerve collection and tissue processing In each subject, a 2 cm segment of the sural nerve was harvested at the site between the lateral malleolus and the Achilles tendon. The side of the limb was randomly chosen except when the injuries were in the biopsy area. In this case, the biopsy was done in the contralateral normal side. The nerve was stretched by hanging with suitable weight in 2.5% glutaraldehyde for 12 min and was then divided into several short segments for better penetration. After this, the nerve was fixed in the same fixative for 12–24 h. The specimens were subsequently washed with 0.1 M cacodylate buffer, pH 7.4, postfixed in 1% osmium tetroxide and dehydrated in increasing concentration of alcohol. Then, the specimens were embedded in Epoxy resin and processed to transverse semi-thin section yielding 1 m-thick slices. Finally, the sections were stained with 1% paraphenylenediamine and prepared for nerve morphometric analysis. 2.3. Morphometric analysis All the sections were screened for artifacts and pathologic appearance under light microscopy prior to the analysis. In each section, the number of fascicles was noted and the area of all fascicles was calculated using Image-Pro Plus software after the images were imported into a microcomputer (under ×4 objective). The total fascicular area did not include the interfascicular zone. As for the count of total fibers, several images under ×40 objective were imported to cover each fascicle. The number of myelinated axons, myelinated axon diameter and myelinated fiber diameter were determined for each fascicle. The diameter of each axon or fiber was derived from the average of 180 values of diameter measured 2◦ apart. g ratio and myelin thickness were then calculated from the corresponding myelinated axon and fiber diameters. These steps were automatically done by the software. However, the results were checked for errors, for example, missed axons or count of small artifacts and appropriate corrections were made by the same investigator. Finally, the
Fig. 1. Micrograph showing the placement of three windows within a fascicle. Normally, two windows were placed in the periphery of fascicle in the opposite direction and one window in the central zone. The area of each window was 0.012 mm2 . Scale bar represents 100 m.
data from all fascicles of each sural nerve section were obtained and the myelinated axon density was calculated by dividing the total number of myelinated axons by the fascicular area. In case of the three-window method, only the fascicles that could contain three full non-adjacent windows under ×40 objective (the area of one window = 0.012 mm2 ) were included. All samples included had at least one fascicle that matched this criterion. Three fields were chosen in the way that they were in a vertical row for each fascicle with two in the periphery and one in the center, as shown in Fig. 1. With this strategy, these three windows were assumed to represent different areas of fascicle in case the non-uniform distribution of nerve fibers was present. The edge of each window must not be in contact with the perineurium or the adjacent window. Axons which were not totally inside the window were discarded as subsequent analysis would yield erroneous values. Similar to the quantification of total fibers, the number of myelinated axons, myelinated axon diameter, myelinated fiber diameter, g ratio, myelin thickness and myelinated fiber density were calculated from the three windows. If there were more than one of such fascicle in each nerve section, these measurement data were derived from all windows. Then, the number of axons was extrapolated for the whole nerve by using the ratio of the total window to the total fascicular areas. Moreover, the frequency distribution of myelinated fiber diameter from all subjects was studied by the count of all fibers and the three-window sampling. 2.4. Statistical analysis To test the accuracy of the three-window sampling method, we examined the degree of agreement between data of the two techniques by determining intraclass correlation coefficient for each measurement parameter. This was done using SPSS for Windows Version 10.0 and single measure intraclass correlation coefficients are shown in Table 1. If this coefficient is close to 1, the agreement between the two techniques is high.
156
V. Chentanez et al. / Journal of Neuroscience Methods 157 (2006) 154–157
Table 1 Comparison of morphometric data of human sural nerve from the quantitation of all fibers and the three-window sampling method Measurement
Count of all fibersa Range
Number of axon
2155–10750
Fiber diameter (m) Axon diameter (m) Myelin thickness (m) g ratio
5.7–8.9 2.2–4.5 1.2–2.8 0.33–0.61
Density of axon (mm−2 )
3585.5–10872.3
Three-window sampling Mean ± S.D.
Range
5672.8 ± 1753.7
2563–11286
± ± ± ±
2.9–9.2 1.4–4.4 1.2–3.3 0.27–0.61
6.9 3.2 1.8 0.48
0.8 0.5 0.3 0.06
6714.2 ± 1560.7
3805.6–10902.8
ICC coefficient Mean ± S.D. 5726.3 ± 1624.6
0.87
± ± ± ±
0.79 0.78 0.88 0.87
7.0 3.1 1.9 0.47
1.0 0.5 0.4 0.07
6843.0 ± 1467.6
0.74
ICC coefficient: single measure intraclass correlation coefficient. a These data have been previously published (Chentanez et al., 2006).
3. Results 3.1. Morphometric data from the quantitation of all myelinated fibers Results of this section were previously published (Chentanez et al., 2006) and summarized in Table 1. They were presented here only for the purpose of comparing with the data derived from the three-window sampling method. The distribution of myelinated fiber diameter from 78 samples is shown in Fig. 2. 3.2. Morphometric data from the three-window sampling method We noticed that the fascicles which could contain three windows of 0.012 mm2 each, in general, had fascicular area more than 0.10 mm2 . Morphometric data are summarized in Table 1. The number of axons in the sural nerve at the ankle level was in the range of 2563–11,286 with an average of 5726.3 ± 1624.6 (mean ± S.D.). The range of myelinated fiber diameter was 2.9–9.2 m and the mean was 7.0 ± 1.0 m. The diameter of myelinated axons ranged from 1.4 to 4.4 m with an average of 3.1 ± 0.5 m. It is worth noting that the lower boundaries of the ranges in the fiber and axon diameters were obviously less than those obtained by the count of all fibers (Table 1). As for myelin thickness, the range was 1.2–3.3 m and the mean was 1.9 ± 0.4 m. Regarding the g ratio, the values ranged from 0.27 to 0.61 with an average of 0.47 ± 0.07. The range of myeli-
nated axon density was 3805.6–10,902.8 mm−2 . The average density was 6843.0 ± 1467.6 mm−2 . The distribution of myelinated fiber diameter from 78 samples is shown in Fig. 2. Bimodal pattern with two peaks at 3–5 and 8–10 m similar to the total count was observed. However, there seems to be more large myelinated fibers when the sampling method was used compared to the count of total fibers. The intraclass correlation (ICC) coefficients of all morphometric parameters were higher than 0.70 (Table 1) with the highest value in the myelin thickness (0.88) and the lowest value in the axon density (0.74). 4. Discussion The ranges and means of all morphometric parameters were similar between the count of total fibers and the three-window sampling method (Table 1). However, it is worth noting that the ranges of the number and density of axons measured by the sampling method were slightly higher relative to those obtained by the total count. Moreover, the lower boundaries of the ranges of fiber and axon diameter were lower when the sampling method was used compared with the count of all fibers. The explanation for these findings might be the sampling of fascicular areas with many small axons, thereby increasing the number of axons along with the axon density and decreasing the lower values of the ranges of fiber and axon diameters. However, these sampling preferences were likely uncommon since the means were still similar between the two techniques. As for the distribution of fiber diameter, similar bimodal pattern was observed with the
Fig. 2. Histogram showing the distribution of myelinated fiber diameters derived from 78 samples. Open and filled bars represent data from the count of total myelinated fibers and the three-window sampling method, respectively.
V. Chentanez et al. / Journal of Neuroscience Methods 157 (2006) 154–157
two techniques. However, there was a tendency toward the larger myelinated fibers when the sampling method was employed compared to the total count. This may be explained by the inclusion of areas with high frequencies of large fibers in the windows and the number was multiplied during the extrapolation to the whole nerve. Nevertheless, the difference was small, 1% or less in each range of diameter. Besides the similarities in means and most ranges as well as fiber diameter distribution, the agreement between the data obtained by the two methods was confirmed by the findings that the ICC coefficients in all parameters were close to 1. This suggests that the three-window sampling method is comparable to the total count and can be used to accurately analyze human sural nerve. Our results were in discordance with the previous studies that showed the inaccuracy of sampling methods by comparing with the results from the total count (Tang and Ebbesson, 1972; Cai et al., 2002). It may be because they used different sampling methods. The selection of every tenth field in the whole nerve cross-section or in one fascicle was employed in those studies. Although this method follows the principle of systematic random sampling, it cannot ensure the inclusion of areas from different parts, for example, center and periphery of nerve or fascicle. The three-window sampling method selects three areas of fascicle, one in the center and two in the periphery on the opposite side to each other. The aim is to make sure that the central and peripheral zones on both sides are taken into account in the analysis. However, the major drawback of this method is that all fascicles in the nerve were not included since only the fascicles large enough to contain three windows of 0.012 mm2 were selected. This also results in the high variation in the percentage of the window areas compared to the total fascicular area among different samples. Nerves with several large fascicles would have higher percentage of the sampling area to the total area in comparison with those containing more small fascicles. Moreover, since morphometric analysis can be done only in large fascicles which can contain three windows, information in the small fascicles is likely to be excluded. As a result, patchy lesions which affect fascicles differently may be missed using this technique. Therefore, whether the three-window sampling can be used to detect pathological changes in the sural nerve with accuracy needs to be verified. In conclusion, the morphometric data derived from the threewindow sampling method were similar to those obtained by the count of total myelinated fibers. Moreover, the agreement between the data measured by these two techniques was confirmed by the high values of intraclass correlation coefficients
157
in all morphometric parameters. As a result, the findings in this study indicate that the three-window sampling technique can be used to obtain representative morphometric data of human sural nerve with less time and effort. However, the application of this technique to diseased nerve remains to be proved. Acknowledgements The authors are indebted to Prof. Giorgio Terenghi, Blond McIndoe Research Laboratories, the University of Manchester, UK, for his invaluable advice regarding the three-window sampling technique. We also would like to thank the Department of Forensic Medicine, especially the Autopsy Unit, for allowing sural nerve biopsy and providing assistance during the procedure. This work was supported by the Ratchadapiseksompoj Research Fund 2003. References Ahmed Z, Brown RA, Ljungberg C, Wiberg M, Terenghi G. Nerve growth factor enhances peripheral nerve regeneration in non-human primates. Scand J Plast Reconstr Surg Hand Surg 1999;33:393–401. Behse F. Morphometric studies on the human sural nerve. Acta Neurol Scand Suppl 1990;132:1–38. Cai Z, Cash K, Thompson PD, Blumbergs PC. Accuracy of sampling methods in morphometric studies of human sural nerves. J Clin Neurosci 2002;9: 181–6. Chentanez V, Cha-oumphol P, Kaewsema A, Agthong S, Huanmanop T. Morphometric data of normal sural nerve in Thai adults. J Med Assoc Thai 2006;89:670–4. Dyck PJ, Sherman WR, Hallcher LM, Service FJ, O’Brien PC, Grina LA, et al. Human diabetic endoneurial sorbitol, fructose, and myo-inositol related to sural nerve morphometry. Ann Neurol 1980;8:590–6. Dyck PJ, Karnes J, O’Brien P, Nukada H, Lais A, Low P. Spatial pattern of nerve fiber abnormality indicative of pathologic mechanisms. Am J Pathol 1984;117:225–38. Dyck PJ, Lais A, Karnes J, O’Brien P, Rizza R. Fiber loss is primary and multifocal in sural nerves in diabetic polyneuropathy. Ann Neurol 1986;19:425– 39. Jacobs JM, Love S. Qualitative and quantitative morphology of human sural nerve at different ages. Brain 1985;108:897–924. Lindemuth R, Ernzerhof C, Schimrigk K. Comparative morphometry of myelinated nerve fibres in the normal and pathologically altered human sural and tibial nerve. Clin Neuropathol 2002;21:29–34. Saxod R, Torch S, Vila A, Laurent A, Stoebner P. The density of myelinated fibres is related to the fascicle diameter in human superficial peroneal nerve. J Neurol Sci 1985;71:49–64. Sterne GD, Brown RA, Green CJ, Terenghi G. Neurotrophin-3 delivered locally via fibronectin mats enhances peripheral nerve regeneration. Eur J Neurosci 1997;9:1388–96. Tang DB, Ebbesson SO. A comparison of a systematic sampling method with complete random sampling for estimating total numbers of nerve fibers. Anat Rec 1972;174:495–502.
Accuracy of the three-window sampling method in morphometric analysis of human sural nerve Vilai Chentanez, Phinidda Cha-oumphol, Athitaya Kaewsema, Sithiporn Agthong ∗ , Thanasil Huanmanop Peripheral Nerve Research Unit, Department of Anatomy, Faculty of Medicine, Chulalongkorn University, Rama IV Road, Pathumwan, Bangkok 10330, Thailand Received 12 January 2006; received in revised form 4 April 2006; accepted 6 April 2006
Abstract Morphometry has an important role in the assessment of sural nerve biopsies as a part of early detection of structural abnormalities in peripheral nerve. Various sampling methods have been used to reduce time and effort needed in the analysis of total nerve fibers but their accuracy remains controversial. We examined the accuracy of three-window sampling method in the morphometric evaluation of human sural nerve biopsies by comparing with the total fiber quantification. Three windows (0.012 mm2 each) were placed in every possible fascicle in the sections and data from all windows were pooled and analyzed for the number of myelinated axons, myelinated fiber diameter, axon diameter, myelin thickness, g ratio as well as myelinated fiber density. Means and ranges of the data from the two techniques were similar and the agreement was further confirmed by intraclass correlation analysis. These findings indicate that the three-window sampling method can be used to evaluate human sural nerve with accuracy. © 2006 Elsevier B.V. All rights reserved. Keywords: Sampling; Sural nerve morphometry
1. Introduction Morphometric analysis of the sural nerve is commonly used to provide more objective and quantitative data compared to conventional visual examination. This is crucial for more accurate and sensitive diagnosis of peripheral nerve diseases. Since it is an energy- and time-consuming procedure to evaluate all the myelinated nerve fibers in the whole section despite an application of computer-assisted image analysis software, various sampling methods have been developed and used (Dyck et al., 1980; Jacobs and Love, 1985; Behse, 1990; Cai et al., 2002; Lindemuth et al., 2002). Systematic sampling is more preferable compared to random and fascicle sampling due to the non-uniform distribution of nerve fibers shown in previous studies (Dyck et al., 1984, 1986; Saxod et al., 1985). However, the accuracy of the sampling techniques to provide valid representative data of the whole myelinated fiber population is still controversial. Tang and Ebbesson (1972) found that approxi-
∗
Corresponding author. Tel.: +66 2 2564281; fax: +66 2 2527028. E-mail address: [email protected] (S. Agthong).
0165-0270/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jneumeth.2006.04.001
mately 50% of the total fascicular area was sufficient to obtain accurate data in cervical sympathetic nerve trunks. Saxod et al. (1985) suggested that the density of myelinated fibers in human superficial peroneal nerve should be calculated by counting all fibers. Cai et al. (2002) proposed that it was necessary to quantify all the myelinated fibers in order to obtain reliable morphometric data of the sural nerve. Despite the above evidence recommending the use of total count, laborious work associated with this technique renders the accurate sampling method still necessary for assessment of the sural nerve. One of the sampling techniques, the three-window sampling, has been used to evaluate nerve regeneration in sciatic nerves of rats and primates (Sterne et al., 1997; Ahmed et al., 1999). The method uses three randomly chosen fields under ×40 objective. However, this sampling technique has not been employed to evaluate human nerves and its accuracy has not been tested. Hence, in this study, we examined the accuracy of the threewindow sampling method compared with the count of total myelinated fibers in human sural nerve in various morphometric parameters and found that the results from both techniques were comparable. As a result, this method can be used to obtain representative morphometric data of human sural nerve.
V. Chentanez et al. / Journal of Neuroscience Methods 157 (2006) 154–157
155
2. Materials and methods 2.1. Subjects Sural nerves were obtained from 78 subjects within 24 h following death. These cases were included with approval from the Department of Forensic Medicine, Faculty of Medicine, Chulalongkorn University. There were 66 males and 12 females and their age range was 20–60 years. This range of age was used due to the report of the mature morphometric details of the sural nerve (Jacobs and Love, 1985). The included subjects must have no known causes or conditions of peripheral neuropathy. Delay duration until removal of nerve after death was 1 h at minimum and 22 h at maximum. The average age of all 78 subjects was 37 years. The most common cause of death was an accident, found in 33 cases (27 males and 6 females). The mean duration after death until the removal of the nerve was approximately 10 h. 2.2. Sural nerve collection and tissue processing In each subject, a 2 cm segment of the sural nerve was harvested at the site between the lateral malleolus and the Achilles tendon. The side of the limb was randomly chosen except when the injuries were in the biopsy area. In this case, the biopsy was done in the contralateral normal side. The nerve was stretched by hanging with suitable weight in 2.5% glutaraldehyde for 12 min and was then divided into several short segments for better penetration. After this, the nerve was fixed in the same fixative for 12–24 h. The specimens were subsequently washed with 0.1 M cacodylate buffer, pH 7.4, postfixed in 1% osmium tetroxide and dehydrated in increasing concentration of alcohol. Then, the specimens were embedded in Epoxy resin and processed to transverse semi-thin section yielding 1 m-thick slices. Finally, the sections were stained with 1% paraphenylenediamine and prepared for nerve morphometric analysis. 2.3. Morphometric analysis All the sections were screened for artifacts and pathologic appearance under light microscopy prior to the analysis. In each section, the number of fascicles was noted and the area of all fascicles was calculated using Image-Pro Plus software after the images were imported into a microcomputer (under ×4 objective). The total fascicular area did not include the interfascicular zone. As for the count of total fibers, several images under ×40 objective were imported to cover each fascicle. The number of myelinated axons, myelinated axon diameter and myelinated fiber diameter were determined for each fascicle. The diameter of each axon or fiber was derived from the average of 180 values of diameter measured 2◦ apart. g ratio and myelin thickness were then calculated from the corresponding myelinated axon and fiber diameters. These steps were automatically done by the software. However, the results were checked for errors, for example, missed axons or count of small artifacts and appropriate corrections were made by the same investigator. Finally, the
Fig. 1. Micrograph showing the placement of three windows within a fascicle. Normally, two windows were placed in the periphery of fascicle in the opposite direction and one window in the central zone. The area of each window was 0.012 mm2 . Scale bar represents 100 m.
data from all fascicles of each sural nerve section were obtained and the myelinated axon density was calculated by dividing the total number of myelinated axons by the fascicular area. In case of the three-window method, only the fascicles that could contain three full non-adjacent windows under ×40 objective (the area of one window = 0.012 mm2 ) were included. All samples included had at least one fascicle that matched this criterion. Three fields were chosen in the way that they were in a vertical row for each fascicle with two in the periphery and one in the center, as shown in Fig. 1. With this strategy, these three windows were assumed to represent different areas of fascicle in case the non-uniform distribution of nerve fibers was present. The edge of each window must not be in contact with the perineurium or the adjacent window. Axons which were not totally inside the window were discarded as subsequent analysis would yield erroneous values. Similar to the quantification of total fibers, the number of myelinated axons, myelinated axon diameter, myelinated fiber diameter, g ratio, myelin thickness and myelinated fiber density were calculated from the three windows. If there were more than one of such fascicle in each nerve section, these measurement data were derived from all windows. Then, the number of axons was extrapolated for the whole nerve by using the ratio of the total window to the total fascicular areas. Moreover, the frequency distribution of myelinated fiber diameter from all subjects was studied by the count of all fibers and the three-window sampling. 2.4. Statistical analysis To test the accuracy of the three-window sampling method, we examined the degree of agreement between data of the two techniques by determining intraclass correlation coefficient for each measurement parameter. This was done using SPSS for Windows Version 10.0 and single measure intraclass correlation coefficients are shown in Table 1. If this coefficient is close to 1, the agreement between the two techniques is high.
156
V. Chentanez et al. / Journal of Neuroscience Methods 157 (2006) 154–157
Table 1 Comparison of morphometric data of human sural nerve from the quantitation of all fibers and the three-window sampling method Measurement
Count of all fibersa Range
Number of axon
2155–10750
Fiber diameter (m) Axon diameter (m) Myelin thickness (m) g ratio
5.7–8.9 2.2–4.5 1.2–2.8 0.33–0.61
Density of axon (mm−2 )
3585.5–10872.3
Three-window sampling Mean ± S.D.
Range
5672.8 ± 1753.7
2563–11286
± ± ± ±
2.9–9.2 1.4–4.4 1.2–3.3 0.27–0.61
6.9 3.2 1.8 0.48
0.8 0.5 0.3 0.06
6714.2 ± 1560.7
3805.6–10902.8
ICC coefficient Mean ± S.D. 5726.3 ± 1624.6
0.87
± ± ± ±
0.79 0.78 0.88 0.87
7.0 3.1 1.9 0.47
1.0 0.5 0.4 0.07
6843.0 ± 1467.6
0.74
ICC coefficient: single measure intraclass correlation coefficient. a These data have been previously published (Chentanez et al., 2006).
3. Results 3.1. Morphometric data from the quantitation of all myelinated fibers Results of this section were previously published (Chentanez et al., 2006) and summarized in Table 1. They were presented here only for the purpose of comparing with the data derived from the three-window sampling method. The distribution of myelinated fiber diameter from 78 samples is shown in Fig. 2. 3.2. Morphometric data from the three-window sampling method We noticed that the fascicles which could contain three windows of 0.012 mm2 each, in general, had fascicular area more than 0.10 mm2 . Morphometric data are summarized in Table 1. The number of axons in the sural nerve at the ankle level was in the range of 2563–11,286 with an average of 5726.3 ± 1624.6 (mean ± S.D.). The range of myelinated fiber diameter was 2.9–9.2 m and the mean was 7.0 ± 1.0 m. The diameter of myelinated axons ranged from 1.4 to 4.4 m with an average of 3.1 ± 0.5 m. It is worth noting that the lower boundaries of the ranges in the fiber and axon diameters were obviously less than those obtained by the count of all fibers (Table 1). As for myelin thickness, the range was 1.2–3.3 m and the mean was 1.9 ± 0.4 m. Regarding the g ratio, the values ranged from 0.27 to 0.61 with an average of 0.47 ± 0.07. The range of myeli-
nated axon density was 3805.6–10,902.8 mm−2 . The average density was 6843.0 ± 1467.6 mm−2 . The distribution of myelinated fiber diameter from 78 samples is shown in Fig. 2. Bimodal pattern with two peaks at 3–5 and 8–10 m similar to the total count was observed. However, there seems to be more large myelinated fibers when the sampling method was used compared to the count of total fibers. The intraclass correlation (ICC) coefficients of all morphometric parameters were higher than 0.70 (Table 1) with the highest value in the myelin thickness (0.88) and the lowest value in the axon density (0.74). 4. Discussion The ranges and means of all morphometric parameters were similar between the count of total fibers and the three-window sampling method (Table 1). However, it is worth noting that the ranges of the number and density of axons measured by the sampling method were slightly higher relative to those obtained by the total count. Moreover, the lower boundaries of the ranges of fiber and axon diameter were lower when the sampling method was used compared with the count of all fibers. The explanation for these findings might be the sampling of fascicular areas with many small axons, thereby increasing the number of axons along with the axon density and decreasing the lower values of the ranges of fiber and axon diameters. However, these sampling preferences were likely uncommon since the means were still similar between the two techniques. As for the distribution of fiber diameter, similar bimodal pattern was observed with the
Fig. 2. Histogram showing the distribution of myelinated fiber diameters derived from 78 samples. Open and filled bars represent data from the count of total myelinated fibers and the three-window sampling method, respectively.
V. Chentanez et al. / Journal of Neuroscience Methods 157 (2006) 154–157
two techniques. However, there was a tendency toward the larger myelinated fibers when the sampling method was employed compared to the total count. This may be explained by the inclusion of areas with high frequencies of large fibers in the windows and the number was multiplied during the extrapolation to the whole nerve. Nevertheless, the difference was small, 1% or less in each range of diameter. Besides the similarities in means and most ranges as well as fiber diameter distribution, the agreement between the data obtained by the two methods was confirmed by the findings that the ICC coefficients in all parameters were close to 1. This suggests that the three-window sampling method is comparable to the total count and can be used to accurately analyze human sural nerve. Our results were in discordance with the previous studies that showed the inaccuracy of sampling methods by comparing with the results from the total count (Tang and Ebbesson, 1972; Cai et al., 2002). It may be because they used different sampling methods. The selection of every tenth field in the whole nerve cross-section or in one fascicle was employed in those studies. Although this method follows the principle of systematic random sampling, it cannot ensure the inclusion of areas from different parts, for example, center and periphery of nerve or fascicle. The three-window sampling method selects three areas of fascicle, one in the center and two in the periphery on the opposite side to each other. The aim is to make sure that the central and peripheral zones on both sides are taken into account in the analysis. However, the major drawback of this method is that all fascicles in the nerve were not included since only the fascicles large enough to contain three windows of 0.012 mm2 were selected. This also results in the high variation in the percentage of the window areas compared to the total fascicular area among different samples. Nerves with several large fascicles would have higher percentage of the sampling area to the total area in comparison with those containing more small fascicles. Moreover, since morphometric analysis can be done only in large fascicles which can contain three windows, information in the small fascicles is likely to be excluded. As a result, patchy lesions which affect fascicles differently may be missed using this technique. Therefore, whether the three-window sampling can be used to detect pathological changes in the sural nerve with accuracy needs to be verified. In conclusion, the morphometric data derived from the threewindow sampling method were similar to those obtained by the count of total myelinated fibers. Moreover, the agreement between the data measured by these two techniques was confirmed by the high values of intraclass correlation coefficients
157
in all morphometric parameters. As a result, the findings in this study indicate that the three-window sampling technique can be used to obtain representative morphometric data of human sural nerve with less time and effort. However, the application of this technique to diseased nerve remains to be proved. Acknowledgements The authors are indebted to Prof. Giorgio Terenghi, Blond McIndoe Research Laboratories, the University of Manchester, UK, for his invaluable advice regarding the three-window sampling technique. We also would like to thank the Department of Forensic Medicine, especially the Autopsy Unit, for allowing sural nerve biopsy and providing assistance during the procedure. This work was supported by the Ratchadapiseksompoj Research Fund 2003. References Ahmed Z, Brown RA, Ljungberg C, Wiberg M, Terenghi G. Nerve growth factor enhances peripheral nerve regeneration in non-human primates. Scand J Plast Reconstr Surg Hand Surg 1999;33:393–401. Behse F. Morphometric studies on the human sural nerve. Acta Neurol Scand Suppl 1990;132:1–38. Cai Z, Cash K, Thompson PD, Blumbergs PC. Accuracy of sampling methods in morphometric studies of human sural nerves. J Clin Neurosci 2002;9: 181–6. Chentanez V, Cha-oumphol P, Kaewsema A, Agthong S, Huanmanop T. Morphometric data of normal sural nerve in Thai adults. J Med Assoc Thai 2006;89:670–4. Dyck PJ, Sherman WR, Hallcher LM, Service FJ, O’Brien PC, Grina LA, et al. Human diabetic endoneurial sorbitol, fructose, and myo-inositol related to sural nerve morphometry. Ann Neurol 1980;8:590–6. Dyck PJ, Karnes J, O’Brien P, Nukada H, Lais A, Low P. Spatial pattern of nerve fiber abnormality indicative of pathologic mechanisms. Am J Pathol 1984;117:225–38. Dyck PJ, Lais A, Karnes J, O’Brien P, Rizza R. Fiber loss is primary and multifocal in sural nerves in diabetic polyneuropathy. Ann Neurol 1986;19:425– 39. Jacobs JM, Love S. Qualitative and quantitative morphology of human sural nerve at different ages. Brain 1985;108:897–924. Lindemuth R, Ernzerhof C, Schimrigk K. Comparative morphometry of myelinated nerve fibres in the normal and pathologically altered human sural and tibial nerve. Clin Neuropathol 2002;21:29–34. Saxod R, Torch S, Vila A, Laurent A, Stoebner P. The density of myelinated fibres is related to the fascicle diameter in human superficial peroneal nerve. J Neurol Sci 1985;71:49–64. Sterne GD, Brown RA, Green CJ, Terenghi G. Neurotrophin-3 delivered locally via fibronectin mats enhances peripheral nerve regeneration. Eur J Neurosci 1997;9:1388–96. Tang DB, Ebbesson SO. A comparison of a systematic sampling method with complete random sampling for estimating total numbers of nerve fibers. Anat Rec 1972;174:495–502.