All animal care and handling was in accordance with the guidelines of the Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC). Animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the Korean Institute of Toxicology (KIT-1609-0339). A total of 34 male cynomolgus monkeys (Macaca fascicularis) of body weight 2.4 kg to 5.5 kg were used in this study. To reduce the chance of mismatches between body weight and brain size, we excluded NHPs whose body weight exceeded 5.5 kg. All CT scans were performed under anesthesia (intramuscular injection of 5 mg/kg ketamine) and all efforts were made to minimize animal suffering.

CT scans were performed with a 16-channel spiral CT scanner (Somatom Emotion, Siemens Healthcare, Munich, Germany). After induction of general anesthesia with an intramuscular injection of ketamine (5 mg/kg), the animal was placed in a prone position, the gantry was aligned with the orbitomeatal line, and the scan was performed with slice thickness of 0.6 mm. After the scan, images were reconstructed into multiplanar reconstruction (MPR) views that visualizes the brain in axial, sagittal, and coronal slices via built-in software. The raw images were exported into the Dicom viewer (RadiAnt DICOM Viewer, Medixant, Poznan, Poland) for further analysis.

To assess the allometric correlation between body weight and endocranial volume, we made longitudinal measurements of the body weight and endocranial volume of all animals. Body weight was measured regularly with a digital scale (Mattler Toledo KCC150, Manualslib, USA) from the age of 35 months up to before CT scanning. Only the animals which showed the normal growth pattern were entered in this study. To measure the endocranial volume in vivo, we exported the CT data from each animal into ImageJ (NIH, Bethesda, MD, USA) and calculated the volume from the vertex to the outer margin of the foramen magnum. Briefly, all brain images were transformed into a ‘stack’ and the scale was set. In each CT slice, a region of interest (ROI) representing the endocranial content was drawn manually. For each slice, the ROI area was multiplied by the slice thickness, and the values for all of the slices were summed to produce the total volume of the brain compartment.

Fig. 1 shows the various skull reference lines used in this study. From the reconstructed 3-D CT images, cranial landmarks including the internal auditory meatus (IAM), porion (PO), anterior clinoid process (ACP), glabella (GL), opisthocranion (OPC), infraorbitale ridge (IOR), tuberculum sellae (TS), ear-bar zero (EBZ), dorsum sellae (DS), foramen magnum (FM), and the inner skull of the vertex bone were identified, then appropriate CT planes were reconstructed so that pairs of cranial landmarks appeared in the same plane. Fifteen connecting lines (skull reference lines) were drawn between pairs of cranial landmarks and their lengths measured with computer software (RadiAnt DICOM Viewer, Medixant, Poznan, Poland). These include the inter-auricular canal line (IAL) between the two internal auricular canals , the interporion line (IPL), the glabella– tuberculum sellae line (GL-TS), the glabella–tuberculum sellae vertical line (GL-TS-VL), the glabella–dorsum sellae line (GL-DS), the glabella–dorsum sellae vertical line (GL-DS-VL), the glabella– opisthocranion line (GL-OPC), the infraorbital ridge–porion line (IOR-PO), the glabella–ear-bar zero line (GL-EBZ), the glabella–ear-bar zero vertical line (GL-EBZ-VL), the tuberculum sellae–ear-bar zero vertical line (TS-EBZ-VL), the anterior clinoid process–ear-bar zero vertical line (ACP-EBZ-VL), the tuberculum sellae–vertex vertical line (TS-VVL), the foramen magnum–vertex vertical line (FM-VVL), and the foramen magnum–tuberculum sellae–vertex vertical line (FM-TS-VVL). Among these lines, the IAL and IPL represent the variability of endocranial volume along the X-axis (Fig. 1A) whereas GL-OPC, GL-EBZ, GL-EBZ-VL, GL-DS, GL-DS-VL, TS-EBZ-VL, GL-TS, GL-TS-VL, and ACP-EBZ-VL represent the variability along the Y-axis (Fig. 1B). TS-VVL, FM-VVL, and FM-TS-VVL represent the variability along the Z-axis (Fig. 1C). The values of these representative skull reference lines were compared with the animal's age, endocranial volume, and body weight to determine the most relevant linear plots.

Fig. 2 shows the process by which we calculated the craniometric indices, which we used to determine accurate stereotactic coordinates in animals whose body weight did not coincide with that of the animals used in the atlases. First, we plotted the length of the skull reference lines against body weight (Fig. 3) and fit the equation y=Kx+B to each plot, where y is the length of the reference line, x is body weight, K is the slope, and B is the y intercept. We then selected the skull reference lines with the highest R² value, to enable the most reliable body-weight-dependent correction along each axis. Next, we used the equation parameters to predict accurate stereotactic coordinates for animals of a given body weight. For example, in determining X-coordinates in 2.5 kg-weighted NHP, two distances (y) can be obtained from the equation of linear regression by inserting the body weights (2.5 and 4.9 kg). After comparing the two distances, we can determine the “craniometric index” to reflect the ratio of different distance in X-axis skull reference line. The craniometric index is >1 if the body weight is bigger than 4.9 kg, and <1 if 4.9 kg. The craniometric index is applied to X-coordinate in 2.5 kg-weighted NHP to determine the final X-coordinates. This procedure is repeated separately for the Y- and Z-axis coordinates to compensate for the non-uniform changes in distance from EBZ with body weight. A similar equation is then applied to the Y axis and the Z axis but with different skull reference lines: GL-EBZ for the Y axis and FM-VVL-TS for the Z axis. The final, adjusted, X, Y, Z coordinates can then be used to approach subcortical targets.

Verification of targeting was performed in three cynomolgus monkeys which were previously euthanized and preserved in compliance with the institutional guidelines of the Korean Institute of Toxicology (KIT). Three animals with body weights of 2.4 kg, 3.0 kg, and 4.1 kg were mounted in the stereotaxic apparatus. Bilateral small trephination holes were made in the skull and, after EBZ zeroing along the X-, Y-, and Z- axes, a microcannula (26 G) was lowered to the stereotactic target and 0.5 ml of methylene blue was infused. The target site in the left hemisphere was derived directly from the stereotactic atlas, whereas the target site in the right hemisphere for corrected by applying the craniometric index to the coordinates. We aimed to target the posterior limb of the internal capsule, between the thalamus and globus pallidus (X=+/−11 mm, Y=13 mm, Z=18 mm from the EBZ point according to the atlas) [1,13]. After dye infusion, the brain was carefully removed from the skull cavity and sectioned to identify the dye stain. Two blind observers measured the distance between dye-staining point in the serially sectioned brain slices and target in stereotactic atlas, and then error distances were averaged in X, Y, Z axes. Two blind observers measured the distance between dye-staining point in the serially sectioned brain slices and target in stereotactic atlas, then error distances were averaged in X, Y, Z axes.

To determine the statistical correlations between age, body weight, endocranial volume, and the skull reference lines, we performed a simple linear regression analysis in ORIGIN 8.5 software (OriginLab, Northampton, MA, USA). This yielded R² values (p<0.001) and a simple regression equation for each variable such as body weight, age, and endocranial volume. We assessed all the correlations between the variables and the different skull reference lines and identified the lines that produced the greatest R² value. An overall F test (p<0.0001) was performed to evaluate the regression model. During validation of the results, we compared the errors (distance from the planned target) using Student's t-test (p<0.05).

Fig. 4 shows the relationships between body weight, age and, endocranial volume for 34 cynomolgus NHPs. All measurements are presented in Table 1. There was an overall increase in body weight with age for body weights between 2.4 kg and 5.5 kg (overall R²=0.73, p<0.0001). Beyond 60 months of age, there was no increase in body weight (Fig. 4A). Fig. 4B and 4C show the relationships between body weight or age and endocranial volume. The correlation score for the relationship between body weight and endocranial volume (R²=0.47) was statistically higher than the correlation score for the relationship between age and endocranial volume (R²=0.30).

Table 2 and Fig. 3 show the relationships between skull reference lines representing the stereotactic X, Y, and Z axes, and age, endocranial volume, and body weight. When the length of skull reference lines was compared with age, endocranial volume, and body weight, the relationship with body weight was the strongest (R²=0.73). Among the skull reference lines, the IPL had the highest R² value of all X-axis lines (R² =0.84, p<0.0001) and the GL-EBZ line had the highest the R² value of all Y-axis lines (R² =0.85, p<0.0001). Among the Z-axis reference lines, the Mag-TS-VVL line had the highest R² value (R²=0.55, p<0.0001). Our results confirm that skull growth is uneven during postnatal development and suggest that a simple brain expansion ratio would be unlikely to reflect the true growth. Skull growth was more prominent in the X and Y axes than the Z axis; consequently, use of skull reference lines that correspond to the different axes is a reasonable approach to determine the individual X, Y, and Z coordinates for the growing brain. Once a skull reference line was selected for each axis, body weight proved to be the most reliable determinant of the length of the reference line from EBZ.

The craniometric index was calculated as the ratio of the reference line length in the atlas to the reference line length derived from the simple linear regression equation. The craniometric equation was calculated separately for each axis, using the reference line with the highest R² value for that axis. Table 3 shows the values used to calculate the craniometric indices for each axis for a 2.5 kg NHP. The craniometric index was similar for the X and Y axes, but larger for the Z axis.

Modification of stereotactic targets by the axis-specific craniometric indices was validated by microinjection of methylene blue. Fig. 5 shows the dye staining after injections for multiple targets for the internal capsule, caudate & putamen and substantia nigra in animals of different body weights. For example, in the NPH with a body weight of 2.4 kg, stereotactic microinjection of methylene blue according to the atlas coordinates (X: 11; Y: 13; Z: 18) stained the upper putamen, outside the internal capsule. We could achieve a similar results in targeting the caudate nucleus, putamen and substantia nigra in 2.4, 3 or 4 kg-weighted NPH (Fig. 5).

Table 4 shows a comparison of the stereotactic coordinates for the right internal capsule in a 2.5 kg NHP derived from the stereotactic atlas, together with the coordinates derived after craniometric index corrections. The X, Y, and Z coordinates were predicted from linear graphs of body weight and each axis was represented by a different skull reference line.

The average errors were 23% (X axis), 12% (Y axis), and 6% (Z axis) in a 2.4-kg NHP, 11% (X axis), 15% (Y axis), and 3% (Z axis) in a 3-kg NHP, and 14% (X axis), 6% (Y axis), and 6% (Z axis) in a 4.3-kg NHP. However, after using the craniometric indices to adjust the stereotactic coordinates, the dye stained the upper portion of the internal capsule correctly in all specimens. The average errors were 5% (X axis), 4% (Y axis), and 11% (Z axis) in the 2.4-kg NHP, 5% (X axis), 2% (Y axis), and 6% (Z axis) in the 3-kg NHP, and 9% (X axis), 6% (Y axis), and 6% (Z axis) in the 4.3-kg NHP.