[1] 23w Changes in volume, surface estimate, three-dimensional shape and total number of neurons of the human primary visual cortex from midgestation until old age

[1] 272w Macroscopic features such as volume, surface estimate, thickness and caudorostral length of the human primary visual cortex (Brodman's area 17) of 46 human brains between midgestation and 93 years were studied by means of camera lucida drawings from serial frontal sections. Individual values were best fitted by a logistic function from midgestation to adulthood and by a regression line between adulthood and old age. Allometric functions were calculated to study developmental relationships between all the features. The three-dimensional shape of area 17 was also reconstructed from the serial sections in 15 cases and correlated with the sequence of morphological events. The sulcal pattern of area 17 begins to develop around 21 weeks of gestation but remains rather simple until birth, while it becomes more convoluted, particularly in the caudal part, during the postnatal period. Until birth, a large increase in cortical thickness (about 83% of its mean adult value) and caudorostral length (69%) produces a moderate increase in cortical volume (31%) and surface estimate (40%) of area 17. After birth, the cortical volume and surface undergo their maximum growth rate, in spite of a rather small increase in cortical thickness and caudorostral length. This is due to the development of the pattern of gyrification within and around the calcarine fissure. All macroscopic features have reached the mean adult value by the end of the first postnatal year. With aging, the only features to undergo significant regression are the cortical surface estimate and the caudorostral length. The total number of neurons in area 17 shows great interindividual variability at all ages. No decrease in the postnatal period or in aging could be demonstrated.

[1] 277w The maturation of human visual function occurs mostly after birth. Physiological features like acuity, contrast sensitivity, selectivity for orientation, colour vision, directional motion, binocularity and spatial attention all develop during the first postnatal year and even during the first postnatal months. They depend mostly on the maturation of cortical operations (Teller and Movshon 1986;Stanley 1991;Atkinson 1992), implying macroscopic and microscopic changes in the visual cortex during the developmental period. Although morphological maturation does not depend only on macroscopic features like volume or surface, the latter give basic information on the time schedule. In particular, the volume of the primary visual cortex represents one of the major features, allowing-together with the estimation of neuronal densities in area 17 from the same brains (Leuba and Garey 1987) -an estimation of the total number of neurons throughout age. This volume seems to reach adult values a few months postnatally (Huttenlocher et al. 1982;Sauer et al. 1983;Huttenlocher and De Courten 1987;Klekamp et al. 1991), but there is no information about how and when its three-dimensional adult shape is formed, in relation to the gyrification pattern and surface of the primary visual cortex. Actually, most of the adult striate cortex is buried within the depth of the calcarine fissure, making it difficult to estimate the surface, which is directly related to the magnification factor in vision. The maturation of the cortical retinotopic map represents an important factor in the maturation of infant vision (Stanley 1991), but surface estimates of the primary visual cortex have not been made throughout development. Aging represents also a critical period for human vision (Kline and Schieber 1985), but its morphological basis has not yet been extensively studied.

[2] 186w In human adults -as in other primates -Brodman's area 17 (the primary visual area 17) is easily distinguishable from adjacent area 18 by the presence of a dense layer 4c and the well-developed stria of Gennari in layer 4b (Le Gros Clark and Sunderland 1939; Fisken et al. 1973), as well as by quantitative cytoarchitectonic (Leuba and Garey 1989) and myeloarchitectonic criteria (Clarke and Miklossy 1990). During maturation, the early development of layers 5 and 6 in area 17 allows a clear demarcation at about 20 weeks of gestation according to Sauer et al. (1983) and at 21 weeks in our cases. We were thus able to draw the outline of area 17 on serial sections in order to measure its volume, thickness and caudorostral length, to estimate its surface and to reconstruct its threedimensional (3D) shape, from midgestation to senescence. This permitted the study of developmental correlations between these macroscopic features. In most cases we had previously measured the neuronal density in area 17 (Leuba and Garey 1987), so that now we could calculate the total number of neurons in area 17 at different ages.

[1] 30w From 21 to 23 weeks of gestation, the limits of primary area 17 with its neighbouring area 18 are already distinguishable due to a lower cellular density in the future

[2] 162w Surface estimate (cm2) 0.8 o~ ........ ~s~ //--~ ...... PR~AT Y~NG 0.7 /~ ....... \\\ 0.6 ~ " .... ----"\ ""... x\xx 0.5 ..... \ / ......... \\ 0.4 7 ......... \\ 0.3 ~ .... \ X /f X ".. \ 0.2 /" ', "'... \ \\ / \ ",, \ j X "'~',,X \ 0.1 / ',,, .... ~, X 0.0 0.08 0.56 1.04 1.52 2.00 2.48 2.96 3.44 3.92 4.40 Caudorostral length (cm) Volume (cm3) 9 8 7 6-5 4a 2i 1 O 4 21W B 8.7M 5 6 7 0 0 0 0 O AGE 20Y 50Y 60Y 70Y 9 80Y 90Y 100Y Surface estimate (cm2) 40 30 20 i 9 i 10 ol ..... // 9 : ..... ZIW B 10M 4 5 6 7 20Y 8 9 AGE 0 0 0 //~-0 -. 0 9 eO 9 "'~"-~. // 50Y 60Y 70Y 80Y 90Y 100Y Fig. 6. Graphic representation of changes of volume and surface estimate (

[3] 306w Table 1) of the primary visual cortex with age. The developmental period is fitted by a logistic growth function (Table 2). Age is represented on a logarithmic scale and numbers given in the second row at bottom are powers to apply to e=2.71, in order to obtain gestational age in days. The first point is indicated at 21 gestational weeks (g0. The vertical lines indicate birth (B) and the age in months (M) where 99% of the plateau value is reached on the curve for each cortical feature. The adult and aging values are fitted by a regression line and represented on a linear scale. The change in the scale of horizontal axis is just before 50 years. Dark circles indicate female brains, open circles male brains; 13 values of cortical volume from Huttenlocher et al. (1982; Table 1) are included 0, 4 Caudo-rostral length (cm) 5 i 9 : 0 4 3 9 ~176 2 1 21W B 6M 5 6 7 20Y 8 9 AGE O 0 50Y 60Y 70Y 80Y 90Y 100Y 359 Thickness (cm) 0"28" ~ rr 0.26: i 9 i 0.24 0 ~ 9 i 0.22 o 0.20 ~ o o.16 0.16 0.14 0.12 0.10 ............................................. // 21W B 7M 20Y 50Y 4 5 6 7 8 9 g•__• infragranular layers 5 6, and from about 30 weeks of gestation the different layers of area 17 are clearly defined (Fig. 1). At 21 weeks of gestation, the 3D shape of area 17 shows the first signs of sulcus formation in the middle of its antero-posterior length, while at 23 25 weeks of gestation, the calcarine sulcus is well formed and appears uniform throughout (Fig. 2a). At 30 weeks of gestation, the sulcal pattern of area 17 remains very simple, although it begins to become more folded at the most posterior level.

[4] 213w In spite of a great increase in volume, this sulcal pattern seems to persist at birth and until about 2 months postnatally (Fig. 2a). From about 3 postnatal months, area 17 exhibits a higher degree of gyrification particularly in its caudal part (Fig. 2b, E). The cortical folding takes place within the calcarine sulcus but at the same time the entire sulcus can be folded on itself towards the caudal pole. In the adults, there are large interindividual variations in the complex 3D shape of area 17, particularly in the most caudal part, which shows a major degree of gyrification in some cases (see Fig. 2b, F and H) but not all (see Fig. 2b, G). Area 17 seems distributed not only in the main calcarine sulcus, but also on side branches of this sulcus as well as sometimes partly on adjacent sulci. However, in all cases there is a medial part along the caudorostral axis where the calcarine fissure presents a deep and regular sulcus without side gyrification. Finally, the most rostral part ending towards the parieto-occipital fissure presents also a less regular structure, being often asymmetrically distributed on both lips. Consequently, the association visual cortex of area 18 may enter partly into the last rostral part of the fissure (Fig. 3).

[5] 333w The changes in complexity of the 3D shape and gyrification pattern of the primary visual cortex are related to the distribution of perimeter-based surface estimates in both development and adult age. In the adults, although there is always a peak of surface distribution in the caudal part of the calcarine fissure, this peak is slightly shifted towards the rostral part. The effect is greater in large volumes (more than 6 cm 3) than in smaller volumes (Fig. 4a, b). The caudorostral extent of the large volumes is also longer, with a greater proportion of the surface estimate in the rostral part (Fig. 4b). If we now relate the surface distribution to the 3D shape of the primary visual cortex, we observe that large volumes of area 17 exhibit a higher degree of complexity and gyrification than smaller ones, with probably a higher degree of extension within the side branches of the calcarine sulcus and within the adjacent sulci (Figs. 2b, 3). During development, the distribution of perimeter based surface estimates of the primary visual cortex along the caudorostral axis of the calcarine fissure exhibits specific features (Fig. 5). In the prenatal period, the distribution is rather uniform, with a decrease in the most rostral part. In the postnatal period, there is a peak in the first caudal part with a "plateau" in the central part and a decrease in the rostral part. However, during this period there are large variations: a brain with a small cortical volume shows a very simple gyrification pattern (Fig. 2a, D) at 2 postnatal months, while a brain with a large cortical volume shows a much more complex pattern (Fig. 2b, E) at almost 3 postnatal months. In the adults, the caudal peak is even larger, and there is also a greater proportion of surface estimate in the rostral part. Finally, in the elderly, the decrease in cortical surface estimate seems to spare the most caudal part but otherwise affects the whole extent of the primary visual cortex.

[6] 15w Volume, surface estimate, thickness and caudorostral length of area 17 (Tables 1, 2; Figs. 6789)

[7] 707w The four characteristic macroscopic features of cortical development follow a similar logistic function, with a rapid growth period followed by a plateau considered as the mean adult value (Figs. 6, 7). However, the rate of growth during pre-and postnatal periods is different for each cortical feature. During prenatal development, the cortical thickness is increasing the most rapidly, reaching 83% of the mean adult value at birth (DMB, i.e. degree of maturity at birth), followed by the caudorostral length (69%), then the surface estimate (40%) and finally the cortical volume (31%) (Table 2). To compare the maturation of volume, surface, thickness and caudorostral length, their plateau value was fitted to 100 (Fig. 8). During postnatal development, the first value to reach maturity, at the time when 99% of P1 is reached, is the caudorostral length (6 postnatal months), then the cortical thickness (7 postnatal months), then the cortical volume (8.7 postnatal months) and finally the cortical surface estimate (10 postnatal months). The maximum growth rate for cortical thickness and caudorostral length occurs before birth, while that for cortical volume and surface estimate occurs in the first postnatal months (Fig. 9). Comparatively, the whole brain weight has a degree of maturity of only 25% at birth and reaches its adult value much later than other features, at 19.7 postnatal months (Fig. 8), with a high growth rate during the whole first // [ i .... +" " i.,' / i,u / ,i i L," ///" i I/ / " ~ // ....................... r .... .......... Leng~ i Sud. :----Thick. i ...... Volume , -....... Weight j/ B 6M 8.7M 19.7M 100 200 300 400 500 600 700 800 900 1000 Age (days) 0.006 0.005 0.004 0.003 0.002 0.001 0.000 100 .... : " ~ t \ .. i t ~ / \ ...... ,: \ , . . . . , . . . . , , .... ~ . . . . , .... , .... B 200 300 400 500 ......... Length Surf. ----Thick. ...... Volume ........ Weight 600 700 800 900 1000 Age (days) 361 Fig. 9. The three-parameters curve was used to represent the growth rate (first derivative of the degree of maturity) for cortical volume, surface estimate, thickness, caudorostral length and brain weight. The vertical line indicates birth (B). The maximum growth rate occurs before birth for cortical thickness and caudorostral length, in the first postnatal months for cortical volume and surface, and during the first postnatal year for the brain weight VOLUME 8.45 6.38 4.31 2.23 0.16 4.8 LENGTH 2.28 7' ........... 7 ''/ ....... ,'" 0.2 .... 7"'-" ...... ,/" 0.232 ~f 0.148 1.39 0.50 0.120 Fig. 10. Graphic representation of the allometric function between cortical volume (cm 3) on one side and cortical thickness (cm) and caudorostral length (cm) on the other side. The resulting curve is represented by vertical lines in the middle of the figure, while the individual values are represented by cubes (males) and pyramids (females). In the first part of the curve (prenatal period), a large increase in thickness or length produces a small increase in volume, while in the second part (postnatal period), the phenomenon is reversed. The volume/ thickness and volume/caudorostral length allometric relationships are represented by vertical lines respectively on the right and front sides of the figure postnatal year (Fig, 9). Allometric function between volume on one side and cortical thickness or caudorostral length on the other side, indicates that there are two different stages of development (Fig. 10). During the first one, a large increase in cortical thickness (ca 0.1 cm) or in caudorostral length (ca 3 cm) results in a small variation in cortical volume (ca 2 cm3), while during the second stage a small variation in cortical thickness (ca 0.02 cm) or in caudorostral length (ca 1 cm) produces a large increase in cortical volume (ca 4 cm3). A similar allometric relationship is observed between the surface estimate and the cortical thickness or the caudorostral length. The first stage grossly corresponds to the prenatal period, while the second corresponds to the postnatal period. In contrast, the allometric function between volume and surface or between cortical thickness and caudorostral length is close to a linear function during both stages (Fig. 10).

[8] 72w The computation and testing of Pearson correlation coefficients demonstrates that the brain weight and volume of area 17 are strongly positively correlated during prenatal development (r=0.952; P=0.0001; n=ll), while after birth, they are no longer significantly correlated. During the prenatal period, the cortical volume is also positively correlated with the cortical thickness (r=0.886; P=0.0007; n=10), with the caudorostral length (r= 0.934; P = 0.0001 ; n = 10) and with the surface

[9] 206w Number of Neurons (X 107) 22 20 18 161 14 i 12i lO 8 9 9 9 9 c/ "~, 9 o 6 ...... 0 9 , ..... ......... ....... // .... 21W B 2Y 20Y 50Y 60Y 70Y 4 5 6 7 8 9 AGE Fig. ll. Graphic representation of the total number of neurons in area 17. The developmental period is fitted with two types of logistic function. Age is represented on a logarithmic scale and numbers given in the second row at bottom are powers to apply to e=2.71, in order to obtain gestational age in days. The vertical line indicates birth. The adult and aging values are fitted by a regression line and represented on a linear scale. The change in the scale of horizontal axis is just before 50 years. Dark circles indicate female brains, open 80Y 90Y 100Y circles male brains. Logistic function using three parameters P1, P2, P3 (Table 2); ------Logistic function using nine parameters: P1 = 18.110; P2= 3.350; P3=0.011; P4= -5.715; P5=6.228; P6=0.014; P7=5.131; P8=-3.896; P9=-0.004; .... Logistic function using nine parameters: P1 = 18.547; P2 = 2.455; P3=0.014; P4=-4.718; P5=2.726; P6=0.012; P7=6.457; P8 = -0.055; P9 = -0.004. See text for details estimate (r=0.980; P=0.0001; n=10

[10] 53w ). In the postnatal period, the cortical volume is mainly correlated with the surface estimate (r = 0.976; P = 0.0008; n = 6) and with the cortical thickness (r=0.874; P=0.0101; n=7). In the adults, the cortical volume is strongly correlated only with the surface estimate (r = 0.971; P=0.0012; n = 6).

[11] 141w During aging (Figs. 6,7), linear regression analysis demonstrates that the slight decrease in brain weight and cortical thickness is not statistically significant. However, the decrease in cortical volume is statistically significant (r=0.54; P=0.0250; n=17), as well as that in surface estimate (r = 0.70; P = 0.0051; n = 14) and in caudorostral length (r=0.75; P=0.0012; n= 15). In order to subtract the influence of brain weight from these variations, we used it as a covariate in a linear regression analysis, which rendered the decrease in cortical volume no longer significant. The decrease in surface estimate became less significant (F(1,10) = 7.47; P = 0.02), as well as that in caudorostral length (F(1,11) = 10.46; P = 0.008). During aging, the volume remained correlated with the surface estimate, although less strongly than during adulthood. (r=0.720; P=0.0287; n=9). 1, 2; Fig. 11)

[12] 73w As for most cases we had estimations both of the volume of area 17 and of its mean neuronal density, we calculated the total number of neurons in area 17 throughout the whole life-span. The dispersion of the values was very high, but data could be fitted using the three-parameter logistic growth function, showing 50% of the adult value (14.142 x 107) reached at 19.3 gestational weeks and 99% at 3.3 postnatal months.

[13] 91w By using the nine-parameter logistic growth function, one can obtain two different shapes exhibiting a more or less pronounced overshoot around birth: the greatest overshoot (about 35%) is obtained by giving more importance in the choice of parameters to data between midgestation and birth, but giving similar importance to data from midgestation to adulthood reveals only a very small postnatal overshoot (8%), with a plateau value similar to that produced with the three-parameter growth function (Fig. 11). Between adulthood and aging, linear regression analysis indicates no statistically significant decrease (Fig. 11).

[14] 377w The classic representation of the visual field on the human striate cortex (Holmes and Lister 1916;Holmes 1917) was recently reviewed (Horton and Hoyt 1991). By correlating magnetic resonance scans with homonymous field defects in patients with clear lesions of the occipital lobe, it was found that the cortical extent devoted to central vision was greater than previously described, 2.5 ~ corresponding to 30%, 10 ~ to 50% and 30 ~ to 83% of the surface of the primary visual cortex. For the calculation of the magnification factor (defined as a unit of striate surface per degree of arc in the visual field), the authors used the mean adult surface estimate of 25 cm 2 given by Stensaas et al. (1974). In our series, this average is 28 cm 2, and in several cases even higher, meaning that the cortical magnification factor may be greater than suggested. Moreover, the representation of the central 10 ~ on the first 50% of the caudorostral length (Horton and Hoyt 1991) is linked to the most convoluted part of area 17 according to our 3D reconstructions. This suggests that the distortion of the visual field representation, particularly that of the horizontal meridian in the bed of the calcarine fissure, could be much greater in the caudal half devoted to central vision. As can also be deduced from our surface estimate distribution (Fig. 5), the regular ellipsoid shape of the flattened striate cortex corresponding to the representation of the visual field (Horton and Hoyt 1991) may not be completely adequate for the representation of central vision in human adults, as is the case in less convoluted primates (Van Essen et al. 1984;Dow et al. 1985) or in the human prenatal period (Fig. 5). During the postnatal period, the distribution of our surface estimate along the caudorostral axis undergoes the greatest changes in the caudal part of the primary visual cortex, linked with accelerated gyrification within and around the calcarine fissure. This supports the notion of rapid maturation for central vision, probably until about 4 postnatal months according to our 3D reconstructions. However the distribution of surface estimate enlarges also in the more rostral part, until the mean adult value is reached at 10 postnatal months, implying also a great maturation of peripheral vision.

[1] 84w We estimated the volume, surface, thickness and caudorostral length of area 17 in 33 brains between midgestation and old age. For the volume only, we included 13 more brains (Table 1) already used in a previous study (Huttenlocher et al. 1982;Huttenlocher and De Courten 1987). Thus 46 human brains of both sexes ranging between 21 weeks of gestation and 93 years were used, none of them having a history of neurological abnormality. Complementary data like brain weight, body weight and height were also recorded.

[2] 265w The brains were weighed, including the cerebellum but not the spinal chord, fixed, and stored in buffered 10% formaldehyde for 1-12 months (post-mortem delay between 6 and 24 h). The whole right occipital lobe was removed, dehydrated and embedded in celloidin. Serial frontal sections of 80 gm were cut, and every fifth was stained with methylene blue. For the purpose of 3D reconstruction, orientation landmarks were cut in 15 brains within the celloidin block and preserved on the serial sections. Before the brains were embedded, shrinkage was measured by the immersion method (Huttenlocher et al. 1982;Huttenlocher and De Courten 1987) and was found to vary with age: 50% between midgestation and birth; 45% between birth and 2 postnatal months; 42% between 2 and 4 postnatal months; 37% between 5 and 8 postnatal months; 35% after 8 postnatal months. From the volumetric shrinkage coefficient, surface and linear shrinkage coefficients were calculated. Raw values of cortical volume, surface estimate, thickness and caudorostral length were corrected. The non corrected volume values were used only for the estimation of the total number of neurons, together with neuronal densities in the same brains, measured on small blocks removed from the middle of the upper bank of the calcarine sulcus and prepared for semithin sections (Leuba and Garey 1987). and became more evident with age (Fig. 1). The volume was then obtained by adding together the cross-sectional areas of individual sections multiplied by five times the section thickness. Parts of area 17 situated in the different branches of the main calcarine sulcus as well as in adjacent sulci were pooled together.

[3] 192w 2. By measuring the cross sectional areas on serial sections of area 17, the perimeters based on the outline of area 17 were also measured. In order to estimate the external surface of area 17 (including side branches and adjacent sulci), the sum of these perimeters was divided by two and multiplied by five times the section thickness, giving an estimate of the surface in a stack of five successive sections at regular points along the caudorostral axis (see 3). Adding together all these values produces a raw estimate of the external surface of area 17. This procedure clearly represents an approximation (Fahle and Palm 1983;Antal et al. 1992), the magnitude of error depending both on the cortical folding and on the orientation of the plane of sectioning in relation to the brain axes. The influence of cortical folding between the successive sections was difficult to take into account (see Discussion), but the plane of sectioning was frontal and similar for all the brains (see inset of Fig. 1). Hence we considered the surface estimate from perimeter as acceptable with regard to the important information about its caudorostral distribution and total extent.

[4] 154w 3. In order to obtain the caudorostral distribution of surface, estimates in a stack of five successive sections were plotted at equidistant points along the caudorostral axis. Each point corresponded to the sagittal distance of the midpoint to the caudal pole (see also 5). In different age groups, mean surface estimates were smoothed with a spline. 4. The cortical thickness of area 17 was estimated from the mean of ten measurements per drawing, on ten drawings at different caudorostrai levels per brain, as the minimum distance between the pia and the limit of area 17 with the white matter, avoiding parts of the sulcus that were grossly enlarged due to oblique cutting. 5. The caudorostral length of area 17 was calculated as the number of serial sections in which area 17 appeared, multiplied by five times the section thickness. The distance of each section from the caudal pole was calculated in a similar way.

[5] 120w 6. Graphic 3D reconstructions were made in 15 cases, using one out of ten sections. For this purpose, the landmarks cut within the celloidin blocks were reported on the drawings obtained with the camera lucida and superimposed on each other when the drawings were fed into a computer using a digitizing tablet connected to a Sigmex graphics terminal and a Vaxstation 3200 computer. A 3D reconstruction program, based on GKS, allowing the removal of hidden lines (Hornung and Kraftsik 1988) generated several angles of view of the stack of recorded sections, which enabled us to visualize the global 3D shape of the primary visual cortex. For the colour reconstructions, we used a program based on surface rendering method (see Acknowledgements).

[6] 88w 7. Values of the uncorrected volume of area 17 were multiplied by the mean neuronal density measured in the same hemisphere when available, in order to calculate the total number of neurons. In most cases, we used the estimations of neuronal numerical density already published (Leuba and Garey 1987), and estimated four additional adult cases with the same method. For the calculation of absolute neuron numbers, the values for volume and neuronal density were used without correction for shrinkage, which was similar in both volume and density measurements.

[7] 77w 1.The volume of area 17 was measured with the basic volume estimator for systematic sampling (Uylings et ai. 1986). Every fifth section was drawn with the aid of a camera lucida device, followed by computer-assisted measurement of cross-sectional area. The limits between areas 17 and 18 were already visible at 21 weeks of gestation, by the presence of a zone of lower cellular density in the inferior cortical plate corresponding to the future layers 5 and 6,

[8] 75w For the growth analysis of all quantitative features, data were fitted by a logistic function showing the lowest sum of squared errors (Kretschmann et al. 1979;Antonelli 1985;Klekamp et al. 1991). The volume, surface estimate, caudorostral length and thickness, as well as the brain weight and absolute neuron number were expressed as functions of age (in days) by the formula: Y (t)=P1/(1 +e(mqP3~~ P1, P2, P3 are parameters to be determined in order to minimize the

[9] 555w Table 1. Quantitative features in the right primary visual cortex at various ages (n = 46) Case Age AgeD Sex Br. weight ~ Volume b Surface ~ Length d Thickness ~ Neur. dens. r Tot. neurons g 353 1 21GW 147 F -0.25 0.794 0.8668 0.1262 --2 23GW 161 F 72 0.30 2.588 0.9172 0.1222 75.450 11.3175 3 25GW 175 F 80 0.22 1.826 0.8164 0.1310 75.510 8.1551 4 25GW 175 M 90 0.16 1.000 0.6652 0.1257 76.840 6.2240 5 28GW 196 F 105 0.46 . . . . . 6 31GW 217 F 162 0.55 4.239 1.2498 0.1350 37.800 10.5840 7 33GW 231 M 200 0.96 6.905 2.1771 0.1460 20.110 9.6528 8 33GW 231 F 260 0.96 6.381 2.4795 0.1821 --9 36GW 252 F 340 2.64 13.144 2.9835 0.1966 14.600 19.2720 10 38GW 266 M 400 2.31 12.081 2.8902 0.1703 9.380 11.9126 11 40GW 280 F 300 1.67 --9.570 8.8044 12 40GW 280 M 368 2.64 16.699 3.0367 0.2421 --13 2PW 294 M -2.42 . . . . . 14 2M 336 M 760 4.95 ---5.180 14.0896 15 2M 336 F 620 2.45 10.785 4.1107 0.2175 --16 2M 336 F 485 3.84 15.239 4.0131 0.2334 --17 2.3M 343 F 550 6.07 26.084 4.5501 0.2359 --18 3M 364 F 550 6.74 24.815 4.1224 0.2601 -19 3M 364 M 610 4.97 20.646 4.3643 0.2203 5.740 16.1294 20 3M 364 M 700 2.18 3.5418 0.2025 --21 4M 392 M 650 6.55 --4.940 18.7720 22 5M 420 M 920 6.10 23.278 3.4630 0.2493 3.830 13.5582 23 8M 504 M 880 6.11 ---3.770 14.5145 24 19M 812 F 1200 6.42 ---2.550 10.6335 25 33M 1204 M 1160 6.82 --3.180 14.0874 26 5Y 2100 M -6.63 ---27 11Y 4284 M 5.57 --2.540 9.1948 28 12Y 4648 F -5.57 . . . . 29 17Y 6468 M 1533 5.03 20.683 3.9804 0.2204 3.180 10.3986 30 26Y 9744 M 1270 5.83 --3.800 14.4020 31 31Y 11564 M 1588 8.45 32.331 4.3498 0.2318 3.650 20.0385 32 48Y 17752 M 1410 6.98 29.652 4.9501 0.2290 3.880 17.6152 33 49Y 18116 M 1340 5.58 23.015 4.5345 0.2367 2.680 9.7284 34 52Y 19208 M 1400 7.16 30.652 4.5807 0.2412 3.140 14.6324 35 55Y 20300 M 1650 8.11 34.197 4.8116 0.2372 2.990 15.7573 36 66Y 24304 F 1110 4.88 19.191 3.2878 0.2345 3.740 11.8558 37 67Y 24668 M 1300 3.75 17.645 4.1190 0.2286 4.130 10.0772 38 71Y 26124 M 1200 5.92 --4.750 18.2875 39 71Y 26124 F 1170 5.83 21.042 4.1651 0.2364 3.850 14.5915 40 73Y 26852 F 890 4.84 21.483 4.2113 0.2401 4.630 14.5845 41 75Y 27580 M -4.47 23.615 3.8881 0.2204 --42 76Y 27944 M 1700 4.42 18.098 3.8882 0.2383 3.710 10.6477 43 81Y 29764 F 970 4.98 21.389 3.6572 0.1949 5.190 16.8156 44 82Y 30128 F 1400 3.25 16.219 3.8885 0.2041 4.570 9.6427 45 85Y 31220 F 1180 5.75 -3.5556 0.2084 3.880 14.5112 46 93Y 34132 F 1320 6.38 22.935 3.2878 0.2400 3.690 15.3135 Volumes in cases 5, 10, 11, 13, 14, 21, 23-28, 30, 38 are taken from Huttenlocher et al. (1982). Our earliest value for cortical volume is at 21 weeks of gestation and that for total neuronal number at 23 weeks of gestation. AgeD, Age in gestational days; GW, gestational weeks; PW, postnatal weeks; M, months; Y, years; -, missing values Brain weight in grams b Corrected volume in cm 3

[10] 147w sum of squares from each data point to the curve (fitting procedure named NLIN (SAS Institute, 1989;1990); t is the age expressed in gestational days and e=2.71 (Table 2). This logistic function produces an S shaped curve, with a plateau value given by P1 (mean adult value), while P2 and P3 influence the shape of the curve. The absolute neuron number was also fitted with a nine parameter logistic growth function proposed by Klekamp et al. (1991) Caudorostral length in cm Thickness in cm f Neuronal density: number per cm 3 of cerebral cortex divided by 107 (Leuba and Garey 1987) g Total number of neurons: uncorrected cortical volume times neuronal density, divided by 10 7 ment. A logarithmic scale was used to represent in more details the developmental period around birth (Figs. 6,7,11), thus making the original points [t,Y (t)] to be represented as [ln(t), Y(t)].

[11] 205w In order to normalize the curves and to compare the maturation of brain weight, cortical volume, surface estimate, caudorostral length and thickness, we used concepts defined mathematically such as the degree of maturity, expressed as a percentage of P1 (mean adult plateau value), and the growth rate calculated as the first derivative of the logistic growth function (Kretschmann et al. 1979;Klekamp et al. 1991). We calculated the degree of maturity at birth (DMB; Table 2), the half value time (HVT; Table 2) Bar = 200 gm. Inset indicates the trace of two frontal sections related to brain axes; the caudorostral distance used for quantification is along horizontal axis A C .. D a E F ~. G H lm Fig. 2a, b. Three-dimensional reconstructions of the right primary visual cortex at various ages. In all the figures, the medial side of the hemisphere i s upper left and the rostral side is down; arrowheads on the reconstructions of the whole volume (left) indicate the levels at which the partial reconstructions shown further to the right, were truncated. Note the high variability in volume and shape in the postnatal period (cases D and E). Bar 200 gin. a A Female, 21 gestational weeks; volume 0.25 cm3

[12] 194w ; B Female, 23 gestational weeks; volume 0.30 cm3; C Female, 33 gestational weeks; volume 0.96 cm3; D Female, 2 postnatal months; volume 2.45 cm 3. b E Female, 2.3 postnatal months; volume 6.07 cm3; F Male, 48 years; volume 6.98 cm3; G Male, 67 years; volume 3.75 cm3; H Female, 81 years; volume 4.98 cm 3 Surface estimate (cm2) 1"01 0.9 0.6 a 357 1,0 " NQ-0.08 0.56 1.04 1.52 2.00 2.48 2.96 Caudorostral length (cm) 3.44 3.92 b Fig. 4. Distribution of the surface estimates of area 17 along the caudorostral axis of the calcarine fissure in a four adult brains with volumes between 3 and 6 cm 3 (cases no 29, 33, 36, 37 of Table 1) and b four adult brains with volumes between 6.1 and 8 cm 3 (cases no 31, 32, 34, 35 of Table 1). The caudal pole is to the left. Each bar represents the mean estimated surface at a defined caudorostral level and each pattern within a bar represents the contribution of the individuals to the mean. Note their homogeneous contribution P1 was reached, and the time of maturity when 99% of P1 was reached.

[13] 45w There is no clear mathematical relationship between the morphometrical features that we have measured, but they are clearly not independent of each other. Thus, allometric relations (Antonelli 1985), based on the previous logistic functions, were calculated between them by means of the following formula :

[14] 150w where Y and Z are two allometric features (for instance volume and thickness or volume and caudorostral Fig. 3a-d. Colour representation of the right primary area 17 surrounded by secondary area 18 (case no 43 of Table 1). Each slice represents a stack of ten serial sections. We used the surface rendering method to show with different colours the superior (violet) and inferior (green) parts of area 18 surrounding the superior (yellow) and inferior (red) parts of area 17 on each side of the calcarine fissure. The extension of the caudorostral axis is 3.66 cm. a View from the caudal pole; the caudorostral axis runs approximately Cases were also divided into groups: prenatal cases (23-40 weeks of gestation), postnatal cases (2 postnatal weeks -12 years), adults (17-55 years) and the elderly (66-93 years) for the analysis of the distribution of surface estimates along the caudorostral extent of the calcarine fissure.

[15] 25w For the analysis of aging, data were fitted with a linear regression function taking into account only the two groups of adults and the elderly.

[16] 52w Pearson correlation coefficients between brain weight, cortical volume, surface estimate, thickness and caudorostral length of area 17 were computed and tested for significance during the prenatal and postnatal period as well as during adulthood and aging (same groups as above) For statistics, we used a statistical analysis package (SAS Institute, 1989, 1990).

[1] 607w We studied the age-related changes of volume, surface estimate, caudorostral length and thickness with a logistic model that has been shown to provide the nearest approximation to the main trends :of morphological brain development, enabling us to use concepts such as degree of maturity and growth rate (Kretschmann et al. 1979;Klekamp et al. 1991). Other authors have studied the volume but did not estimate the surface extent of developing area 17 (Huttenlocher et al. 1982;Sauer et al. 1983;Murphy 1985;Klekamp et al. 1991). All their values are in the same range as ours, except those of Murphy (1985), which were calculated without a shrinkage coefficient. In the series of 13 cases of Huttenlocher et al. (1982), an adult value was already reached around 4 postnatal months. Fitting their data with the three-parameter logistic growth function, we found similar values. In the series of Sauer et al. (1983), data were fitted by a logistic growth function which gave a plateau at 8 postnatal months. In the series of male brains of Klekamp et al. (1991), data of Sauer et al. (1983) were included and the nine-parameter logistic growth function was used, revealing an overshoot of cortical volume with a maximum at 8 postnatal months and a subsequent decline to adult level. In our series of male and female brains, including also the data of Huttenlocher et al. (1982), the postnatal variability was very high, masking the overshoot if any. The degree of maturity at birth in our cases is only 30% of the adult value, while that of the cases of Klekamp et al. (1991) is already 50%, but the adult level is reached around 8 postnatal months in both series. Our maximum value (8.45 cm 3) in the adults is from male brains only, while in development and aging they are from both male and female brains. This may introduce a small bias as Klekamp et al. (1989) have shown that males generally have larger brain volumes than females. The external surface of area 17 has not been estimated during the developmental period in human brains. As a matter of fact, the surface of a 3D structure is not a trivial parameter to measure, and most methods introduce an error, smaller when based on the triangulation algorithm (Fahle and Palm 1983). The use of the latter supposes precise conditions that were not fulfilled in all our cases. For instance, the exact alignment of successive sections was available only in the 3D reconstructions. Moreover, the cortical shape was changing very rapidly; sometimes, parts of area 17 could not be found from one section to the next, preventing the use of corresponding points for the triangulation method. However, estimates of the surface of area 17 based on the sum of the perimeter measurements multiplied by the total length of the structure have been widely used in primates (Van Essen et al. 1984;Missler et al. 1993;Purves and Lamantia 1993), including adult human beings (Stensaas et al. 1974). The comparison of adult human estimates by various authors gave values between 20 cm 2 (Putnam 1926;Stensaas et al. 1974) and 34 cm 2 (Brodman 1918). We made an estimate of 28 cm 2 for the adult, with a wide range between 15 and 40 cm 2, which is in agreement with others. The high interindividual adult variability of the surface estimate and 3D shape in our cases confirms the variability in the amount and distribution of the primary visual cortex exposed on the borders or buried inside the calcarine fissure (Stensaas et al. 1974). Such a high interindividual variability has also been reported in the human lateral geniculate nucleus (Hickey and Guillery 1979).

[2] 237w As morphological features like volume, surface estimate, cortical thickness and caudorostral length are obviously not independent of each other, the allometric function (Antonelli 1985) was most suitable for studying their relationships during development. A first prenatal stage could be defined, when a large variation in cortical thickness or caudorostral length produced a small variation in cortical volume or surface estimate, and a second postnatal stage when it was the opposite. In contrast, cortical volume and surface, as well as cortical thickness and caudorostral length, remained almost linearly linked during the whole period. Further, the study of Pearson correlations between these morphological features confirmed these allometric relationships. Thus, the acceleration in the growth of volume and surface estimate after birth, without corresponding growth in cortical thickness or caudorostral length has to be explained by the gyrification and folding process around the occipital pole and within the calcarine sulcus and its side branches. As a matter of fact, a high index of gyrification has been described in the occipital part of the adult human brain (Zilles et al. 1988). However, although adult-like patterns of the 3D shape of the primary visual cortex are already observed around 3 postnatal months, the maturation of quantitative features of the human primary visual cortex is completed only by the end of the first year, with the estimate of the external surface being the last to reach the adult value at 10 postnatal months.

[3] 168w With aging, the cortical grey matter is known to shrink, particularly in the frontal cortex (Haug et al. 1983). We found a statistically significant decrease in the volume of area 17 with age, and this difference remained if we analysed male brains separately across adulthood and aging to avoid the bias mentioned earlier. However, by using the brain weight as a covariate to subtract both the sex effect and the secular effect leading to an increase in brain weight for successive generations (Haug 1985), this difference became insignificant. We did not find any further decrease in the volume of area 17 in pathological conditions such as Alzheimer's disease (Leuba and Kraftsik 1994). Among the different parameters contributing to the volume of area 17, only the caudorostral length and the surface estimate decrease significantly with aging, even when we take into account the secular effect. This means that there may be a loss of functional columns in area 17 with aging, or a regular size reduction of the columns.

[4] 278w Taking into account our data of neuronal density in area 17 of the same brains (Leuba and Garey 1987;1989), we have now been able to estimate the total number of neurons and found a mean of about 140 millions in adult area 17, for the right hemisphere. Our estimation is in the same range, although slightly higher, as that of Klekamp et al. (1991), who found about 112 millions in the right hemisphere. However, our data exhibit a twofold range in the total number of neurons, from 90 to 200 millions. Similarly, a range in the total number of retinal ganglion cells and optic nerve axons, going from 700,000 to 2 millions, was reported in human adults by Curcio and Allen (1990) in their own as well as in data of others. Altogether, these data support the existence of a real interindividual variability in the number of ganglion cells and axons as well as of cortical neurons. As the interindividual variability in the cortical neuronal density of our cases was never so high (Leuba and Garey 1987;1989), it is clearly the volume of the primary visual cortex that makes the difference, with a twofold range. Such a large range in the cortical volume (Sauer et al. 1983;Murphy 1985) or surface (Stensaas et al. 1974) was also reported in human and macaque monkey brains (Van Essen et al. 1984), but its functional significance is still a matter of discussion. It would be particularly interesting to know if individual neuron numbers in area 17 correlate with retinal ganglion cell numbers and if individuals with a larger cortical volume and more neurons in area 17 are better suited to visual tasks.

[5] 439w During development, ganglion cell death has been shown in the human fetal retina between 14 and 30 weeks of gestation (Provis 1987), and elimination of retinal axons occurred at about the same period, reaching an adult-like value of 1.1 million at 29 weeks of gestation (Provis et al. 1985). In the lateral geniculate nucleus, data on cell elimination are available only in the rhesus mon-key (Williams and Rakic 1988), demonstrating that cell death occurs before the establishment of geniculo-cortical connections and of geniculate cell layers, and before the depletion of retinal axons, being probably a key factor controlling the number of connections. In the primary visual cortex of the rhesus monkey, O'Kusky and Colonnier (1982) estimated a loss in total number of neurons of about 15% in the postnatal period. In our present estimation of the total number of neurons in area 17 through age, the mathematical model with the smallest asymptotic error (Marquardt 1963) is the one fitted with the three-parameter logistic function that shows a plateau value at 3.3 postnatal months without any overshoot in the postnatal period. By using the nine-parameter logistic function, data may be fitted with curves of different shapes, depending on how the underlying model and the initial parameters are chosen. However, among a defined number of solutions, only two are useful for our developmental model. They show either a very small overshoot (about 8%), when a similar importance is given to data between 23 weeks of gestation and adulthood in the choice of parameters, or a greater overshoot (about 35 %) when parameters are chosen giving a greater importance to the period around birth. In this last situation however, the asymptotic error is also greater. Finally, as there are relatively few values within a large range in the perinatal period, our data can neither prove nor exclude that the overshooting model corresponds to biological reality. Klekamp et al. (1991), using our values of neuronal density (Leuba and Garey 1987) together with the values of Sauer et al. (1983) for the volume of area 17, have calculated a 31% postnatal decrease in total neuronal number by using a decreasing three-parameter logistic function. Using a similar function applied to the postnatal period, we found only a 17% decrease in total neuronal number. This difference is probably due to the fact that our value for cortical volume at birth represents only 30% of the adult value, while that of Sauer et al. (1983) is already 50%. As we checked that estimation of gestational age was correct by following the curves of body weight and size, these differences have to be accounted for by interindividual variability.

[6] 103w Between adulthood and aging, the slight decrease shown by the regression line is not statistically significant, meaning that there is no significant neuronal death in normal aging of the visual cortex. This was also suggested by the data of Haug et al. (1984) and Haug (1985), but not by those of Devaney and Johnson (1980) obtained with a cell dispersion technique. In fact, the decrease in cortical volume with normal aging is compensated for by an increase in neuronal density (Leuba and Garey 1987), which is not the case in Alzheimer's disease, where we found a real neuronal loss (Leuba and Kraftsik 1994).

[7] 253w In relation with the columnar organization of the human visual cortex (Horton and Hedley-White 1984;Horton et al. 1990), the increase in cortical surface estimate until 10 postnatal months suggests that the postnatal maturation of the primary visual cortex is essentially due to an increasing columnar width owing to neuropile (Takashima et al. 1980;Becker et al. 1984;Michel and Garey 1984;Zilles et al. 1986), synaptic (Huttenlocher et al. 1982;Huttenlocher and De Courten 1987) and axonal development. This in turn should be linked with the maturation of the retinotopic map, and with that of the cortical magnification factor (Stanley 1991). Strikingly, several physiological characteristics mature early in the postnatal period (Atkinson 1984;1992), in particular visual acuity and contrast sensitivity between birth and 6 months of age, and colour vision and selectivity for orientation and directional motion in the first postnatal months. Binocular function does not emerge before 3 months postnatally and matures during the first year (Atkinson 1984;1992;Mohn and Van Hof-Van Duin 1986). Thus, the first year, and even the first 4-8 postnatal months appear as particularly important for the correlative maturation of anatomical and physiological features in the human infant. However, fine mechanisms of maturation, for instance the pattern of neurofilament immunoreactivity (Ang et al. 1991) or that of cytochrome oxidase activity (Wong-Riley et al. 1993) may still develop later. Also, a process of synapse elimination contributing to the fine tuning of connections could occur around the time of puberty, as shown in the primary visual cortex of the macaque monkey (Bourgeois and Rakic 1993).

[8] 196w During aging, the incidence of visual dysfunctions such as decrease in acuity and stereoacuity, dark adaptation and colour discrimination is greater (Kline and Schieber 1985). Although these perturbations could result from changes in the eye or retina, their association with cortical deterioration cannot be ruled out. The decrease of surface estimate in the primary visual cortex seems to occur more in its anterior part than in the posterior one, matching the fact that central vision tends to be preserved in normal aging (Kline and Schieber 1985). In spite of this, we could not demonstrate any loss of neurons in area 17 with aging, meaning that it is probably the neuropile that is progressively damaged. A loss of myelin in the stria of Gennari has been reported with aging (Lintl and Braak 1983), as well as dendritic changes in some regions, although not specifically in the visual area (Coleman and Flood 1987). However, apart from regressive events, compensatory growth phenomena have also been reported at the dendritic and synaptic level, preserving a functional threshold (Coleman and Buell 1985;Bertoni-Freddari et al. 1990). This may well be the situation in the human primary visual cortex under normal aging conditions.