Neuroscience Letters 304 (2001) 93±96
Unbiased stereological quanti®cation of neurons in the human spiral ganglion Akira Ishiyama a,*, Jody Agena a, Ivan Lopez a, Yong Tang a,b a
Division of Head and Neck Surgery, UCLA School of Medicine, Los Angeles, CA 90095, USA b Department of Neurology, UCLA School of Medicine, Los Angeles, CA 90095, USA
Received 2 February 2001; received in revised form 23 March 2001; accepted 23 March 2001
Abstract We applied an unbiased stereological technique, the optical fractionator, on ®ve human archival temporal bone specimens to estimate the total number of spiral ganglion neurons. Available archival human temporal bone specimen has been serially sectioned at 20 mm and every tenth section was stained. All the stained sections passing through the spiral ganglion were used for the analysis. From each section sampled, the counting areas were systematically randomly sampled within the sectional area of the spiral ganglion. The neurons within the counting areas sampled were counted with the optical disector technique. The total number of the human spiral ganglion neurons was estimated by multiplying the number of neurons counted by the reciprocal of the aggregate sampling fraction. We found an average of 41 700 neurons with a coef®cient of variation of 0.14, which is a signi®cant departure from the previously published data obtained with the assumption-based methods. The mean coef®cient of error for the stereological estimates of the total number of human spiral ganglion neurons was 0.078. The present report presents unbiased stereological sampling and counting strategies for the future quantitative studies on the spiral ganglion neurons. The result of the present study provides the ®rst unbiased baseline value of the human spiral ganglion neurons. q 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Human temporal bone; Spiral ganglion; Neuron; Number; Stereology, Fractionator
Spiral ganglion neurons are the primary afferent neurons that transmit auditory information from the organ of Corti to the cochlear nuclei. The number of neurons in the normal human spiral ganglion is important for evaluating changes associated with aging or pathological conditions. There have been a few studies on the number of spiral ganglion neurons, but the results are con¯icting [6,9,11]. Methodologically, the total number of spiral ganglion neurons has previously been estimated by counting the number of neuronal pro®les and then applying the correction factors [12,13] to the pro®le counts [6,9,11]. However, the correction factors are based on the assumption that the possibility of spiral ganglion neurons appearing at the interface between sections is a constant (5 or 10%). Unbiased stereological methods use a three-dimensional probe, the disector, to sample and count objects (cells, nucleoli, synapses, etc.) without relying on any assumptions about the sizes, shapes, and orientations of the objects to be counted . Numerous * Corresponding author. Tel.: 11-310-206-2041; fax: 11-310794-5089. E-mail address: [email protected]
studies have already demonstrated remarkable discrepancies between results obtained from traditional assumptionbased methods and unbiased stereology techniques [8,10,17]. Therefore, we undertook a study to obtain an unbiased estimate of the total number of spiral ganglion neurons from ®ve temporal bone specimens with no documented hearing loss at Victor Goodhill Ear Center archival temporal bone collections at UCLA. Stereological analysis was accomplished with the assistance of the Stereo Investigator v 3.0 software (MicroBrightField, Inc., Colchester, VT). A personal computer running the software and a monitor were connected to a color video camera mounted at the top of a Nikon Optiphot microscope. The motorized stage (LEP Bio-point) of the microscope controlled by the software allows for precise well-de®ned movements along the x- and y-axes. A Heidenhain microcator was attached to the stage for precise measurement of focal depth. High image resolution and a thin focal plane is obtained using a high numerical aperture (NA 1.40) oil immersion objective lens. Five specimens from the patients with no evidence of
0304-3940/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S03 04 - 394 0( 0 1) 01 77 4- 8
A. Ishiyama et al. / Neuroscience Letters 304 (2001) 93±96
hearing loss were chosen from the archival temporal bone collections at the Victor Goodhill Ear Center at UCLA. Medial records and autopsy reports revealed no diseases or treatments that might have affected the auditory system. According to the fractionator principle , systematic random sampling of the tissues from human spiral ganglion is accomplished by systematically randomly sampling a known fraction of the sections that pass through the spiral ganglion, a known fraction of the sectional area in the sampled sections, and a known fraction of the section thickness in the sampled areas. Following removal at autopsy, the temporal bones were ®xed in formalin solution, decalci®ed with EDTA, embedded in celloidin, and sectioned in the axial (horizontal) plane at 20-mm-thickness. Every tenth section was stained with hematoxylin and eosin. The sections passing
through spiral ganglion were used for the analysis, which means that the section sampling fractions (ssf) is 1/10. Because sections were cut without knowing the spiral ganglion's position within the temporal bone, the ®rst section examined can be considered as a randomly starting section in accordance to the systematic random sampling rule of the fractionator . A systematic random sample of the sectional area of spiral ganglion from each sampled section was achieved by positioning an unbiased counting frame  of known area on the section with constant intervals in the x and y-axes. In practice, in each of the sections sampled, a contour was traced around spiral ganglion under 1 £ (Fig. 1A). Using the motorized stage controlled by Stero Investigator software, a sampling grid was randomly placed over the contour (Fig. 1B). Each box of the sampling grid contains an unbiased
Fig. 1. (A): A contour has been traced around the spiral ganglion. (B): View of the computer-generated sampling grid that is placed over the spiral ganglion in a random fashion. A rectangular counting frame (two-dimensional unbiased counting frame) is seen at the upper left corner of each box of the sampling grid. (C,D): Counting of nucleoli. Focal depth in mm (measured by the microcator attached to the microscope stage) is indicated on the focal depth bar, where 0 mm represents the top surface of the section. Negative depth values indicate that the microscope is focusing downwards into the section. An unbiased counting frame is put on (C) and (D). The thick border of the frame and its extension is the exclusion line, and the thin border of the frame is the inclusion line. The counting rule of the optical disector is that nucleoli that come into focus within the height of the optical disector and within the counting frame are counted when one moves the focal plane continuously through the section. A nucleolus is considered to be within the counting frame if it is entirely within the counting frame or partially within the counting frame without touching the exclusion line of the counting frame when it ®rst comes into focus. (C): Focal depth is at 27.20 mm. (D): Focusing downward to 28.30 mm brings one nucleolus indicated by arrow into focus. Therefore, one nucleolus is counted. Scale bars 1000 mm in (A), 50 mm in (B), 5 mm in (C).
A. Ishiyama et al. / Neuroscience Letters 304 (2001) 93±96
counting frame, representing the area in which sampling will occur. The x and y dimensions of each box in the sampling grid represent the x and y distance displacing one counting frame to the next. As illustrated in Fig. 1B, the area of each counting frame takes up a fraction of the area of each box in the sampling grid. This fraction represents the area sampling fraction (asf). The random placement of the sampling grid ensures the random positioning of the ®rst counting frame to be analyzed. In the present study, the height and width of the counting frame were set at 30 mm and 40 mm, respectively, and the x and y dimensions of each box in the sampling grid were set at 120 mm and 120 mm, respectively. Hence, the area sampling fraction, asf, was (30 mm£40 mm)/(120 mm£120 mm) 1200 mm 2 /14400 mm 2 1/12. In the present study, the actual section thickness was measured at the ®ve positions selected at uniform intervals within the section area of spiral ganglion by a Heidenhain microcator (MT12, Germany) attached to the stage of the microscope. After all sections were examined, the overall average section thickness of the specimen was calculated from the average thickness of the individual sections. A few mm guard areas at the top and bottom surfaces of the section were assured to avoid physical disruptions and lost caps at the surfaces of the section. Therefore, the section thickness sampling fraction (stsf) equals to the height of the tissue used for neuron counting (h) divided by the actual section thickness (t). In the present study, h was set at 10 mm, and t was 21 mm. Therefore, stsf h/t 10/21. In each of the counting areas sampled from spiral ganglion, the number of neurons was counted with the optical disector principle . In the present study, the ®rst focal plane of the disector (the top of the disector) was positioned about 5 mm below the upper surface of the section. Counting was performed through a depth of 10 mm (the height of the disector), leaving approximately 5 mm at the bottom of the section. When focusing through a section, several criteria were used to distinguish between neurons and glial cells . The speci®c unit chosen for counting was the nucleolus. An unbiased counting frame  was superimposed on the tissue image viewed on the monitor by the Stereo Investigator software (Fig. 1C,D). The counting rule of the optical disector is that within the height of the optical disector, nucleoli that come into focus are counted if they are entirely within the counting frame or partially within the counting frame without touching the exclusion lines of the counting frame (Fig. 1C,D). Once every section had been analyzed, the total number, N, of nucleoli (and hence neurons) in the entire specimen was calculated by multiplying the total number of nucleoli actually counted, SQ 2 , in the sampled volume of tissue, with the reciprocal of all three sampling fractions mentioned above . The equation is: N SQ 2 £ 1=ssf £ 1=asf £ 1=stsf. The coef®cient of error (CE) for the stereological estimation of the neuronal number in human spiral ganglion was calculated according to Gundersen et al. (1999) . This study found an average of 41 700 (coef®cient of
variation (CV) 0.14) spiral ganglion neurons in the ®ve normal specimens studied (Table 1). On average, 160 (ranging from 128 to 189) neurons were counted per specimen, which provided a mean coef®cient of error (CE) of 0.078 for the individual stereological estimates of the total number of neurons. With respect to design ef®ciency, it may be noted, as shown in Table 1, that the biological variance constituted 69% of the total variance. This indicates that the precision of the individual stereological estimate is satisfactory since the major factor contributing to the total interindividual variability was the biological variability of the individuals studied. Our results differ signi®cantly from the results of the previous studies [6,9,11]. Previous studies on the number of the human spiral ganglion neurons have produced a wide range of results (from 23 910 to 38 352 neurons) [6,9,11]. Age is an important variable in the number estimates of neurons but age differences can explain only small part of the variability in those estimates. Previous studies counted all the spiral ganglion cells that showed a clear nucleolus in every tenth section with thickness of 20 mm. In fact, the authors counted the pro®les of ganglion cells rather than the ganglion cells. The total number of spiral ganglion cells was estimated by multiplying the number of neuronal pro®les counted by 10, and then by a factor of 0.9  or 0.95  to account for doubly counted cells located at the interface between sections [6,9,11]. The correction factors used by previous studies are based on assumptions that every neuron within each spiral ganglion has the same shape, size and orientation and that they all have a 5 or 10% probability of appearing on two sections with the section thickness of 20 mm when they are ten sections apart. However, those assumptions are almost never met since the human spiral ganglion contains neurons with variable sizes, shapes, and orientations. In addition, unlike Table 1 The stereological estimates of the number of the neurons in the ®ve spiral ganglions a Patient no.
1 2 3 4 5 Mean SD CV 1 ± CE 2/CV 2
22 22 22 52 60 35.6 18.8 0.53
M M F F M
45 47 33 43 38 41 5
CE* 700 100 600 900 000 700 700 0.14
0.073 0.073 0.087 0.080 0.077 0.078 0.69
The total number of the neurons in human spiral ganglion, N; standard deviation of the mean, SD; coef®cient of variation, CV (CV SD/Mean); contribution to the total variance from biological variance, 1 - CE 2/CV 2 . *CE is the coef®cient of error for the stereological estimation of the neuronal number in human spiral ganglion. The way of estimating CE for the observations obtained from systematically randomly sampled sections taken along one axis was described by Gundersen et al. .
A. Ishiyama et al. / Neuroscience Letters 304 (2001) 93±96
unbiased stereological methods in which the number of neurons is counted in three-dimensional space , the previous methods counted the number of neuronal pro®les in which nucleolus was visible in two-dimensions (the surface of section). Given the thickness of temporal bone section (20 mm), it is highly likely that small nucleoli in the deep part of the section would escape detection, leading to an erroneous underestimation. Furthermore, small nucleolar fragments at the top and bottom surfaces of a section could also escape detection because of lost caps, i.e. small nucleolar fragments unnoticed or missing at the surfaces of the section. Failure to account for these fragments would lead to further underestimation. The most important feature of the unbiased stereological method used in this report is that neurons are directly counted with the three-dimensional probe, the disector . Moreover, the estimation is essentially independent of the lost caps and unevenness of the section surface due to the use of a few micrometer guard zones at the top and bottom surfaces of a section. Currently available human temporal bone specimens have been serially sectioned at 20 mm according to the traditional protocol described by Schuknecht . For future morphometric studies on human temporal bone specimens thicker sections (i.e. 25 mm or thicker) are recommended. In order to avoid physical disruptions and lost caps at the surfaces of a section, sometime a larger guard zones close to both surfaces of a section are required . The optical fractionator method involves in sampling a known fraction of the tissue from the region of interest and counting all the neurons inside the sampled tissue using the optical disector. The total number of neurons in the region under study is obtained from the number of neurons counted multiplying by the reciprocal of the aggregate sampling fraction and the optical fractionator is unaffected by tissue shrinkage or expansion that takes place during any stage of the preparation of the tissue. The results of neuronal number estimated with optical fractionator are, therefore, independent of dimensional changes in the tissue. This enables the analysis of frozen, vibrotome, and paraf®n sections, in which shrinkage during the staining or embedding stage is signi®cant and dif®cult to measure . In conclusion, we applied the optical fractionator to estimate the total number of spiral ganglion neurons for the ®rst time. The unbiased stereological method can be applied to a large number of archival human temporal bone specimens available to estimate the normative data of spiral ganglion neurons in different age groups. Thus, this method would be invaluable for future studies, including quanti®cation of the spiral ganglion neurons in aging and pathological conditions.
Grant Sponsor: National Institutes of Health Grants DC 00140±02 and AG 09693  Andersen, B.B. and Gundersen, H.J.G., Pronounced loss of cell nuclei and anisotropic deformation of thick sections, J. Microsc., 196 (1999) 69±73.  Gundersen, H.J.G., Notes on the estimation of the numerical density of arbitrary pro®les: The edge effect, J. Microsc., 111 (1977) 219±223.  Gundersen, H.J.G., Stereology of arbitrary particles. A review of unbiased number and size estimators and the presentation of some new ones, in memory of William R. Thompson, J. Microsc., 143 (1986) 3±45.  Gundersen, H.J.G., Bagger, P., Bendtsen, T.F., Evans, S.M., Korbo, L., Marcussen, N., Mùller, A., Nielsen, K., Nyengaard, J.R., Pakkenberg, B., Sùrensen, F.B., Vesterby, A. and West, M.J., The new stereological tools: disector, fractionator, nucleator and point sampled intercepts and their use in pathological research and diagnosis, APMIS, 96 (1988) 857±881.  Gundersen, H.J.G., Jensen, E.B., Kieu, K. and Nielsen, J., The ef®ciency of systematic sampling in stereology±reconsidered, J. Microsc., 193 (1999) 199±211.  Hinojosa, P., Seligsohn, R. and Lerner, S.A., Ganglion cell counts in the cochleae of patients with normal audiograms, Acta Otolaryngol (Stockh.), 99 (1985) 8±13.  Kroustrup, J.P. and Gundersen, H.J.G., Sampling problems in a heterogeneous organ: quantitation of relative and total volume of pancreatic islets by light microscopy, J. Microsc., 132 (1983) 43±55.  Morrison, J.H. and Hof, P.R., Life and death of neurons in the aging brain, Science, 278 (1997) 412±419.  Otte, J., Schuknecht, H.F. and Kerr, A.G., Ganglion cell populations in normal and pathologica human cochleae. Implications for cochlear implantation, Laryngoscope, 88 (1978) 1231±1246.  Pakkenberg, B. and Gundersen, H.J.G., Neocortical neuron number in humans: effect of sex and age, J. Comp. Neurol., 384 (1997) 312±320.  Pollak, A., Felix, H. and Schrott, A., Methodological aspects of quantitative study of spiral ganglion cells, Acta Otolaryngol (Stockh.), 436 (1987) 37±42.  Schuknecht, H.F., Techniques for study of cochlear function and pathology in experimental animals, Arch. Otolaryngol., 58 (1953) 377±397.  Schuknecht, H.F., Neuroanatomical correlates of auditory sensitivity and pitch discrimination in the cat, In G.L. Rasmussen and W. Windle (Eds.), Neural Mechanisms of the Auditory and Vestibular Systems, Charles C. Thomas, Spring®eld, IL, 1960, pp. 76±90.  Schuknecht, H.F., Methods of removal, preparation, and study, In H.F. Schuknecht (Ed.), Pathology of the Ear, 2nd Edition, Lea and Febiger, Philadelphia, PA, 1993, pp. 1±3.  Sterio, D.C., The unbiased estimation of number and sizes of arbitrary particles using the disector, J. Microsc., 134 (1984) 127±136.  West, M.J., Slomianka, L. and Gundersen, H.J.G., Unbiased stereological estimation of the total number of neurons in the subdivisions of the rat hippocampus using the optical fractionator, Anat. Rec., 231 (1991) 482±497.  West, M.J., Regionally speci®c loss of neurons in the aging human hippocampus, Neurobiol. Aging, 14 (1993) 287±293.