I didnt have time to read this, but i was wondering if it of any relevance (i know its about women but wutever).
Does it have anything to do with why some hairs are thinner at the root.
Summary
Shed head-hair fibres of young (1620-year-old), nonalopecic women (n = 25), exhibiting both exogen clubs and anagen tips (EA) were studied. Such fibres are shown, for the first time, to comprise 44% of shed hair and to form a uni-modal, positively skewed distribution with a mean length of 16.7 ± 4.9 cm, which is also correlated with the length of the haircut. As individual fibres exit the skin in early anagen VI, their major-axis diameters increase rapidly to maxima at about 25% of their total potential length and subsequently decrease to their exogen clubs, at a rate of 1.31% per cm (n = 28). EA diameters are further correlated with their lengths. Maximal and proximal diameters increase by 1.40% per cm and 1.02% per cm increments in fibre lengths, respectively (P < 0.0001 each; n = 14), these changes being also different from each other (P < 0.001). Besides identifying and characterizing a new class of normal hair (EA) which will probably feature prominently in future hair research, this study reveals several other important aspects of hair growth: (i) the classically described concept of hair miniaturization in androgenetic alopecia (Androgenetic Alopecia) is excessively broad and should therefore be revised; (ii) female Androgenetic Alopecia need not necessarily require a mechanism for rapid miniaturization as recently proposed; and (iii) the putative large variability of normal hair diameters is significantly overestimated, which further opens the field of hair diameter evaluation as a biological indicator of disease and physiological function.
Introduction Go to: Choose Top of page Introduction << Materials and methods Results Discussion References
Human head-hair is usually truncated by the haircut, so that the natural length it may attain is obscured. It is usually estimated to range between 25 and 90 cm, using anagen (growth) phase durations of 27 years and constant hair growth rates of a little over 1 cm per month. These estimates, however, apply to human hair in the abstract, and there do not seem to be any studies in which anagen durations were compared for individual normal subjects.
While the naturally large estimated range of hair lengths would ordinarily imply that normal anagen durations for individuals vary greatly as well, 1 it is often assumed to the contrary, that within-head anagen durations are essentially uniform. Careful consideration of this assumption suggests that it is mandated, at least in part, by the concept of miniaturization. Miniaturization classically describes the pathological mechanism of alopecia androgenetica (Androgenetic Alopecia), in which particular affected hair follicles produce progressively thinner and shorter fibres through successive growth cycles. 14 After several such cycles (each consisting of chronologically ordered anagen, catagen, telogen, exogen 5,6 and latency 7,8 phases; the latter being optional), only the barely visible vellus fibres of Androgenetic Alopecia are produced. It is evident from this definition that any short fibre in a particular hair sample could be considered as 'miniaturizing'. Thus, miniaturization would 'appropriate' short hair for itself, and by default define only fibres of maximal (and hence uniform) lengths, as 'normal'.
Given the importance of miniaturization for diagnosis and treatment of hair loss it should be important to distinguish between miniaturizing, and small, but otherwise normal fibres. The existence of some such fibres is evidenced by the small lanugo-like fibres that are found at the periphery of the hair mass, such as the nape of the neck. However, the relevance of these fibres toward the much more numerous 'terminal' fibres of the main hair mass is undetermined. Studies by the author (unpublished) on hair of different lengths, either shed and cut at their distal ends (EC) or proximally clipped (from the occiput) and displaying their distal anagen initiation tips (CA), have shown him that both the maximal diameters, occurring near anagen tips, 9 and the proximal diameters, near exogen clubs, are correlated with fibre lengths. This suggested that fibres exhibiting both the proximal exogen clubs and the distal anagen tips (EA), should occur frequently in shed hair, and that their diameters are likewise correlated with their lengths. Because EA represent entire anagen growth phases, analysis of their lengths and diameters could conceivably yield valuable information with respect to normal hair growth as differentiated from that during miniaturization. Such an analysis is presented in this report, which for the first time characterizes EA. Shed hairs were obtained from young, nonalopecic women, who could not reasonably be supposed in stages of either Androgenetic Alopecia, or 'early onset' Androgenetic Alopecia. The EA characteristics of these donors are therefore properly assigned to hair in the normal (nonalopecic) state.
Materials and methods Go to: Choose Top of page Introduction Materials and methods << Results Discussion References
Hair samples and fibre length determinations
Exogen fibre samples were obtained from 26 randomly selected, young (1620-year-old) women donors, undergoing routine hair treatments at a local hair dressing school. Each sample was collected onto a clean brush by brushing the entire scalp evenly. One of the donors apparently suffered from telogen effluvium (by her account, and this was confirmed by > 600 shed fibres) and her hair sample was not used further. All of the women were Caucasian and none had evident hair loss (type 0 on the Ludwig scale). 10
Exogen hair clubs can be felt by passing the hair between two fingers. Being discoloured, they can also be visualized against a dark background. Tapering anagen tips can be similarly visualized by holding the hair against a background of contrasting colour. A great majority of the fibres of most donors exhibited the exogen club; however, in two cases a substantial number of fibres cut both proximally and distally (CC) were also found. These were found in particularly curly hair that were also frequently knotted, thus suggesting their generation by mechanical damage. 11 EA and EC were counted and sorted into 5-cm length-classes.
Major-axis diameter determinations of hair-groups along their shafts
Preparations of EA for group-diameter measurements consisted of aligning and taping fibres, previously sorted into 5-cm length classes, from both their proximal and distal ends and at subsequent regular intervals (Fig. 1a). Only hairs longer than 10 cm were prepared in this way, which excluded presumptive miniaturized or peripheral hair as described above. Overall major-axis diameters of the taped hair-groups were measured using a dedicated device which has been described previously. 12,13 Briefly, the hair groups are placed in a window of the device which can close in two dimensions while its edges are kept on a single geometrical plane (the third dimension). Closing the window in the first dimension orients the fibres single-file with the major-axes (of their generally elliptical cross sections), in parallel to the floor of the window, and closing the window in the second dimension brings the fibres together. Their average major-axis diameter is obtained from the width of the window floor when the fibres just cover it, divided by their number. The device is mounted on a micrometer which is used to measure the width of the window with a precision of 1 µm, and the hair are measured along their shafts at locations 2 mm removed from the edges of the tape-tabs (yielding in this case measurements at 1-cm intervals, by using 0.6-cm-wide tape tabs spaced every 2 cm). A dissecting-microscope (40 magnification) is used to determine the end point of the measurements, which is easily identified by a concerted tilting of the fibres when they begin to be compressed in the device window. A coefficient of variation of 1% has been calculated for repeated measurements of fibres at the same places (n > 100).
Statistics and calculations
The SPSS program for statistical analysis was used when possible. 14 All statistical tests are parametric and two-tailed unless otherwise specified and P < 0.05 are taken to be significantly different from expectation. Results are given as mean ± SD.
To convert mixed EC and EA length histograms (of individual donors) into single 'interpolated/extrapolated' EA histograms (Fig. 2), the following four-step algorithm was used: (i) EC length classes as long or longer than the longest EA were pooled and reclassified as a new 'longest EA' class [ EA k+1, where k is the number of EA length classes]; (ii) starting with the shortest EC length class and going to the longest, their fibres were distributed into EA classes longer than them by multiplying their numbers [n(EC) i , where i represents the i th EC length class] by EA relative frequencies [f(EA) i+j, where f(EA) i+j = n(EA) i+j /sum{n(EA) i+j to k+1}, j = 1 to k + 1 i] and the resulting numbers were added to new 'interpolated' EA classes [new n(EA i+j ) = old n(EA i+j ) + n(EC) i *f(EA) i+j ]. This process was repeated recursively until all EC fibres were thus disposed of; (iii) the interpolated (and observed) EA histograms generally presented as positively skewed distributions. An exponential decay curve [y =y 0 + Ae(xx0 )/t , where x0 is the distance from the scalp to the mode of the distribution, A is the number of fibres at the mode, x is the distance from the scalp, y 0 is the y axis asymptote and t is the decay constant] was fitted to the interpolated data from its peak distally; (iv) the EA k+1 class, which had accumulated a considerable number of new fibres during the recursive part of the procedure, was partitioned into new 'extrapolated' EA length classes, calculated using the exponential decay equation from the previous step, and until all the fibres it contained (to the nearest length class), were exhausted.
Results Go to: Choose Top of page Introduction Materials and methods Results << Discussion References
EA and EC length comparisons
Table 1 lists EA and EC counts of the different donors, their mean lengths and standard deviations, and probability values that the lengths of the two types of fibres represent the same populations. Mean EA counts per donor amount to 44% ± 20% of the total exogen hair, and their mean lengths (16.7 ± 4.9 cm) to 59% ± 12% the EC lengths. A significant correlation between EA relative frequencies and lengths of the haircut (represented by mean EC lengths) was expected, and found, but was rather weak (r = 0.41, P < 0.05, one tail). A much stronger correlation (r = 0.77, P < 0.0001), between EA and EC mean lengths, was also found.
EA length histograms generally describe uni-modal positive skew distributions with modes at 7.517.5 cm. EC hair histograms interpenetrate those of EA, but here fibre numbers increase with length and the distribution is usually abruptly truncated at its longer end. These features are illustrated in Fig. 2 (left panel) using data of the three donors with the largest numbers of EA (W, X and Y). On the assumption that EC represent the same EA uni-modal distributions (though truncated by the haircut) as those observed, they were converted to putative EA (Fig. 2, right panel) by the algorithm described above (see Materials and methods). The resulting extrapolated EA (longer than the haircut) distributions were thereby extended to lengths of 77.5, 87.5 and 127.5 cm for samples W, X and Y, respectively.
Major-axis diameter changes along (within) and between EA
As previously shown for CA, 9 EA major-axis diameters similarly increase from their proximal clubs to a maximum near their distal ends, after which they sharply decrease. The maximal hair diameter can therefore be used as a boundary to distinguish the anagen tip, distally, from the hair shaft 'body', proximally. Fig. 1b and c illustrates how total (l), shaft-body (lm) and anagen tip (lt) lengths, and proximal (dp) and maximal (dm) diameters were defined for hair groups and how their rates of increases within and between fibres were calculated. Numbers of fibres and their measured data, for long and short (') hair groups are listed in Table 2. These are in fact just the two extreme hair-group lengths, out of several (numbers in parentheses next to donors) that were usually measured for each donor; these only are listed in the table to allow direct comparisons between all of the donors. Rates of shaft body diameter increases toward their maxima (1.03 ± 0.52 µm/cm and 1.20 ± 0.54 µm/cm, sw and sw', respectively) are linear in general (data not shown) and not significantly different. Their combined rate of increase is 1.11 ± 0.53 µm/cm, or 1.31 ± 0.72%/cm.
Estimates of hair diameter changes between fibres (Table 2, last two columns) used proximal and maximal diameters as natural reference points. Proximal diameter increases (sp) between short and long fibres average 0.84 ± 0.62 µm/cm (1.02 ± 0.76%/cm) and those of maximal (sm) diameters 1.44 ± 0.67 µm/cm (1.40 ± 0.66%/cm). These are significantly different from 0 (P < 0.0001, binomial distribution) and from each other (P < 0.001, Student's t-test, paired samples).
Regression analysis of proximal and maximal diameter changes between fibres of the same donors with six or more measured EA length classes (Table 2, numbers in parentheses for M, U, W and Y) are consistent with linear increases (Fig. 3). The correlations between maximal diameters and hair lengths are stronger than those for proximal diameters, consistent with the overall larger sm as opposed to sp values of Table 2. Diameter changes between fibres as determined by linear regression slopes (Fig. 3) are correlated with their respective sp and sm values (Table 2) for the same donors (r =0.97, P < 0.01, Spearman rank correlation). This implies that the sp and sm values, which were calculated using just the two extreme hair groups, are fairly robust estimates.
Anagen tip vs. EA length correlation
Anagen tip lengths (llm for each hair group in Table 2), were found to be linearly correlated (Fig. 4) with EA lengths (r = 0.73, P < 0.0001, n = 28). Their mean length, as derived from the slope of the regression, subsumes 22% of the mean EA lengths. A similar plot, using anagen tip lengths of all of the hair groups measured (n = 65, sum of numbers in parentheses in Table 2) yields a stronger correlation (r = 0.83), with the tips representing 27% of the whole fibre lengths.
Discussion Go to: Choose Top of page Introduction Materials and methods Results Discussion << References
In a previous work on hair diameters along their shafts, Hutchinson and Thompson, 9 measured major- and minor-axes diameters of EC and CA (called telogen and anagen hairs, respectively) and by collating regions of overlapping diameters were able to produce diameter profiles of presumptive EA (full-length hairs). These authors found proximally decreasing major-axis diameters, although to a lesser extent than those described here (0.45%/cm vs. 1.31%/cm). However, they did not recognize the existence of EA of different lengths, and consequently the correlations between EA lengths and their diameters and anagen tip lengths were not described. With hindsight, the results of this study show that the methodology of collating different EC and CA to produce composite EA profiles is likely to conceal length differences between EA and thus perpetuates a misconception that they are basically of the same length. A large EC and a small CA, for example, may have matching diameters after relatively long and short anagen phase durations, respectively, rather than after the same anagen durations as thought. 9 This will lead to a single composite EA length that is intermediate between their actual disparate ones. Moreover, as anagen durations can be arbitrarily redefined by shifting to locations of overlapping diameters, most combinations of similar EC and CA will converge to apparently uniform EA.
EC, which constitute 56% of shed hair and generally represent EA longer than the haircut were used here to extrapolate EA length distributions (Fig. 2). While the algorithm used is admittedly crude, it does seem adequate as a first approximation. On the one hand it does not make any assumptions with regards to the interpolated distributions, other than that they preserve the observed EA distributions, and on the other hand extrapolation of the EA histograms past their peaks by an exponential decay function seems qualitatively justified by the data, has a precedent (in ageing), 15 and at any rate approaches linearity for longer fibres, where it obtains practical expression. The maximal EA lengths that were obtained (77.5, 87.5 and 127.5 cm) for the three samples evaluated seem substantially longer than analogous values obtained by Hutchinson and Thompson (49 cm), which is understandable (see above). But they also suggest that EA are longer than generally accepted (2590 cm). Interestingly, several lines of investigation support an idea that hair is in fact substantially longer than currently believed. In a study of the EA length distribution of one older woman, without hair loss and who had never cut her hair, this author (unpublished) demonstrated a uni-modal EA length distribution extending to 125 cm. Similarly long EA, as determined by anagen durations of up to 11 years, were also found in one (and to date perhaps the only) well documented study, which followed individual (ageing) follicles over time. 8 Furthermore, a recent phenomenological study of women with long hair 16 concluded that hair lengths up to 183 cm fall within the linear range of a plot of hair lengths vs. numbers of observed women. In contrast, however, the strong correlation between mean EA and EC lengths as described here suggests that hair styles are not arbitrary, but that women 'perceive' the peak lengths of their EA distributions and cut their hair to approximately two to three times this length. Long anagen tips ( 25% of the total EA length as found here), would become increasingly more obvious in longer haircuts and may thus assist in that perception. Further investigations would be required to resolve these matters. Proximal diameter measurements of EA and (individually measured) EC could probably settle the problem of completely describing the length distribution of a hair mass despite the haircut.
The use of EA longer than 10 cm for diameter measurements effectively excluded putative miniaturized fibres, or small nonminiaturized fibres found at the hair mass periphery, from further consideration and thus the correlations between EA diameters and their length apply to terminal fibres. A similar correlation has been described previously (together with positively skewed EA length distributions described as log-normal) in male Androgenetic Alopecia and ageing 7,8, 15 but was attributed to miniaturization. It would be unreasonable to suppose that the correlation found here is also due to miniaturization for several reasons. In the first place, the broad distributions of lengths and diameters that were found would imply that if miniaturization was the responsible factor it would have started several cycles previously, which would be impossible, given the young age of the donors. 17 A supposition of early onset Androgenetic Alopecia would be likewise rejected, and further ruled out on the grounds that the prevalence of EA in shed hair (significant subpopulations in 22/25 donors) would make this the normal, rather than the rare condition it is currently thought to be. 3,18 Finally, as hair diameter normally increases to young adulthood, possibly even to the third decade, 1921 an assumption of 'normally occurring early onset' Androgenetic Alopecia would be incongruous, as it would imply that hair diameter and length are negatively correlated in young women, and was not found to be the case. Thus, it seems inescapable that EA of different lengths represent normal fibres under natural growth conditions.
Because the correlation between normal EA diameters and lengths would make such fibres indistinguishable from miniaturizing hair, it is tempting to speculate that Androgenetic Alopecia may represent a shift of the normal EA distribution toward shorter fibres. Such a shift would appear to move the focus in Androgenetic Alopecia from individually affected follicles to regions of the scalp or perhaps to the entire scalp. The patterned hair loss in male Androgenetic Alopecia appears to gratuitously support this idea, and it would include hair loss in ageing as a natural rather than pathological phenomenon, which seems preferable. It is, however, toward understanding female Androgenetic Alopecia that this concept could be most usefully applied at this time.
Although classical miniaturization was originally described in female Androgenetic Alopecia, where bimodal length and diameter distributions were observed, 2,3 recent studies 21,22 failed to demonstrate a (strong) correlation between hair density and diameter decreases in this condition. A rapid type of miniaturization has therefore been proposed to explain this apparent anomaly in female Androgenetic Alopecia. 17,21 In its most extreme form 17 rapid miniaturization is postulated to take place and be completed between catagen of one follicular cycle and (early) anagen I of the next. It further stipulates that hair diameter decreases (or any diameter changes for that matter) cannot occur during hair growth out of the scalp (anagen VI) on the grounds that this is determined by rapidly amplifying (matrix) cells derived only once per follicular cycle from stem cells at the bulge region. 23 The timing of the switch from normal to miniaturized follicles specifically dissociates observable hair growth from diameter changes and thus circumvents the problem of differentiating between miniaturizing and normal hair. However, no positive evidence has been produced to demonstrate rapid miniaturization so far, and the assumption that diameter changes do not occur during anagen VI (though possibly not critical for rapid-miniaturization, 21,22 ) is clearly contradicted by this, and other studies. 9,12, 13 Thus, it should be regarded as speculative at this time.
EA of differing lengths (the existence of which rapid miniaturization would paradoxically confirm as normal), could possibly explain the anomalous results found in female Androgenetic Alopecia without requiring rapid miniaturization. Slow miniaturization, operating on the entire scalp, could set up a cascade of EA lengths that might easily last over several follicular cycles. The mean diameters of the cascade, being correlated with fibre lengths as they are, would however, remain relatively steady. The shortest hair would eventually become lost (unobserved) and thus apparent hair density would decrease without concomitant diameter decreases. An exception to the cascade, which may be of low significance (and would remain mostly latent so long as the hair is cut), would be found with the largest fibres. Interestingly, preferential diameter decreases of large fibres in female Androgenetic Alopecia have been described. 21
EA, which have not been previously recognized as a class of hair
Does it have anything to do with why some hairs are thinner at the root.
Summary
Shed head-hair fibres of young (1620-year-old), nonalopecic women (n = 25), exhibiting both exogen clubs and anagen tips (EA) were studied. Such fibres are shown, for the first time, to comprise 44% of shed hair and to form a uni-modal, positively skewed distribution with a mean length of 16.7 ± 4.9 cm, which is also correlated with the length of the haircut. As individual fibres exit the skin in early anagen VI, their major-axis diameters increase rapidly to maxima at about 25% of their total potential length and subsequently decrease to their exogen clubs, at a rate of 1.31% per cm (n = 28). EA diameters are further correlated with their lengths. Maximal and proximal diameters increase by 1.40% per cm and 1.02% per cm increments in fibre lengths, respectively (P < 0.0001 each; n = 14), these changes being also different from each other (P < 0.001). Besides identifying and characterizing a new class of normal hair (EA) which will probably feature prominently in future hair research, this study reveals several other important aspects of hair growth: (i) the classically described concept of hair miniaturization in androgenetic alopecia (Androgenetic Alopecia) is excessively broad and should therefore be revised; (ii) female Androgenetic Alopecia need not necessarily require a mechanism for rapid miniaturization as recently proposed; and (iii) the putative large variability of normal hair diameters is significantly overestimated, which further opens the field of hair diameter evaluation as a biological indicator of disease and physiological function.
Introduction Go to: Choose Top of page Introduction << Materials and methods Results Discussion References
Human head-hair is usually truncated by the haircut, so that the natural length it may attain is obscured. It is usually estimated to range between 25 and 90 cm, using anagen (growth) phase durations of 27 years and constant hair growth rates of a little over 1 cm per month. These estimates, however, apply to human hair in the abstract, and there do not seem to be any studies in which anagen durations were compared for individual normal subjects.
While the naturally large estimated range of hair lengths would ordinarily imply that normal anagen durations for individuals vary greatly as well, 1 it is often assumed to the contrary, that within-head anagen durations are essentially uniform. Careful consideration of this assumption suggests that it is mandated, at least in part, by the concept of miniaturization. Miniaturization classically describes the pathological mechanism of alopecia androgenetica (Androgenetic Alopecia), in which particular affected hair follicles produce progressively thinner and shorter fibres through successive growth cycles. 14 After several such cycles (each consisting of chronologically ordered anagen, catagen, telogen, exogen 5,6 and latency 7,8 phases; the latter being optional), only the barely visible vellus fibres of Androgenetic Alopecia are produced. It is evident from this definition that any short fibre in a particular hair sample could be considered as 'miniaturizing'. Thus, miniaturization would 'appropriate' short hair for itself, and by default define only fibres of maximal (and hence uniform) lengths, as 'normal'.
Given the importance of miniaturization for diagnosis and treatment of hair loss it should be important to distinguish between miniaturizing, and small, but otherwise normal fibres. The existence of some such fibres is evidenced by the small lanugo-like fibres that are found at the periphery of the hair mass, such as the nape of the neck. However, the relevance of these fibres toward the much more numerous 'terminal' fibres of the main hair mass is undetermined. Studies by the author (unpublished) on hair of different lengths, either shed and cut at their distal ends (EC) or proximally clipped (from the occiput) and displaying their distal anagen initiation tips (CA), have shown him that both the maximal diameters, occurring near anagen tips, 9 and the proximal diameters, near exogen clubs, are correlated with fibre lengths. This suggested that fibres exhibiting both the proximal exogen clubs and the distal anagen tips (EA), should occur frequently in shed hair, and that their diameters are likewise correlated with their lengths. Because EA represent entire anagen growth phases, analysis of their lengths and diameters could conceivably yield valuable information with respect to normal hair growth as differentiated from that during miniaturization. Such an analysis is presented in this report, which for the first time characterizes EA. Shed hairs were obtained from young, nonalopecic women, who could not reasonably be supposed in stages of either Androgenetic Alopecia, or 'early onset' Androgenetic Alopecia. The EA characteristics of these donors are therefore properly assigned to hair in the normal (nonalopecic) state.
Materials and methods Go to: Choose Top of page Introduction Materials and methods << Results Discussion References
Hair samples and fibre length determinations
Exogen fibre samples were obtained from 26 randomly selected, young (1620-year-old) women donors, undergoing routine hair treatments at a local hair dressing school. Each sample was collected onto a clean brush by brushing the entire scalp evenly. One of the donors apparently suffered from telogen effluvium (by her account, and this was confirmed by > 600 shed fibres) and her hair sample was not used further. All of the women were Caucasian and none had evident hair loss (type 0 on the Ludwig scale). 10
Exogen hair clubs can be felt by passing the hair between two fingers. Being discoloured, they can also be visualized against a dark background. Tapering anagen tips can be similarly visualized by holding the hair against a background of contrasting colour. A great majority of the fibres of most donors exhibited the exogen club; however, in two cases a substantial number of fibres cut both proximally and distally (CC) were also found. These were found in particularly curly hair that were also frequently knotted, thus suggesting their generation by mechanical damage. 11 EA and EC were counted and sorted into 5-cm length-classes.
Major-axis diameter determinations of hair-groups along their shafts
Preparations of EA for group-diameter measurements consisted of aligning and taping fibres, previously sorted into 5-cm length classes, from both their proximal and distal ends and at subsequent regular intervals (Fig. 1a). Only hairs longer than 10 cm were prepared in this way, which excluded presumptive miniaturized or peripheral hair as described above. Overall major-axis diameters of the taped hair-groups were measured using a dedicated device which has been described previously. 12,13 Briefly, the hair groups are placed in a window of the device which can close in two dimensions while its edges are kept on a single geometrical plane (the third dimension). Closing the window in the first dimension orients the fibres single-file with the major-axes (of their generally elliptical cross sections), in parallel to the floor of the window, and closing the window in the second dimension brings the fibres together. Their average major-axis diameter is obtained from the width of the window floor when the fibres just cover it, divided by their number. The device is mounted on a micrometer which is used to measure the width of the window with a precision of 1 µm, and the hair are measured along their shafts at locations 2 mm removed from the edges of the tape-tabs (yielding in this case measurements at 1-cm intervals, by using 0.6-cm-wide tape tabs spaced every 2 cm). A dissecting-microscope (40 magnification) is used to determine the end point of the measurements, which is easily identified by a concerted tilting of the fibres when they begin to be compressed in the device window. A coefficient of variation of 1% has been calculated for repeated measurements of fibres at the same places (n > 100).
Statistics and calculations
The SPSS program for statistical analysis was used when possible. 14 All statistical tests are parametric and two-tailed unless otherwise specified and P < 0.05 are taken to be significantly different from expectation. Results are given as mean ± SD.
To convert mixed EC and EA length histograms (of individual donors) into single 'interpolated/extrapolated' EA histograms (Fig. 2), the following four-step algorithm was used: (i) EC length classes as long or longer than the longest EA were pooled and reclassified as a new 'longest EA' class [ EA k+1, where k is the number of EA length classes]; (ii) starting with the shortest EC length class and going to the longest, their fibres were distributed into EA classes longer than them by multiplying their numbers [n(EC) i , where i represents the i th EC length class] by EA relative frequencies [f(EA) i+j, where f(EA) i+j = n(EA) i+j /sum{n(EA) i+j to k+1}, j = 1 to k + 1 i] and the resulting numbers were added to new 'interpolated' EA classes [new n(EA i+j ) = old n(EA i+j ) + n(EC) i *f(EA) i+j ]. This process was repeated recursively until all EC fibres were thus disposed of; (iii) the interpolated (and observed) EA histograms generally presented as positively skewed distributions. An exponential decay curve [y =y 0 + Ae(xx0 )/t , where x0 is the distance from the scalp to the mode of the distribution, A is the number of fibres at the mode, x is the distance from the scalp, y 0 is the y axis asymptote and t is the decay constant] was fitted to the interpolated data from its peak distally; (iv) the EA k+1 class, which had accumulated a considerable number of new fibres during the recursive part of the procedure, was partitioned into new 'extrapolated' EA length classes, calculated using the exponential decay equation from the previous step, and until all the fibres it contained (to the nearest length class), were exhausted.
Results Go to: Choose Top of page Introduction Materials and methods Results << Discussion References
EA and EC length comparisons
Table 1 lists EA and EC counts of the different donors, their mean lengths and standard deviations, and probability values that the lengths of the two types of fibres represent the same populations. Mean EA counts per donor amount to 44% ± 20% of the total exogen hair, and their mean lengths (16.7 ± 4.9 cm) to 59% ± 12% the EC lengths. A significant correlation between EA relative frequencies and lengths of the haircut (represented by mean EC lengths) was expected, and found, but was rather weak (r = 0.41, P < 0.05, one tail). A much stronger correlation (r = 0.77, P < 0.0001), between EA and EC mean lengths, was also found.
EA length histograms generally describe uni-modal positive skew distributions with modes at 7.517.5 cm. EC hair histograms interpenetrate those of EA, but here fibre numbers increase with length and the distribution is usually abruptly truncated at its longer end. These features are illustrated in Fig. 2 (left panel) using data of the three donors with the largest numbers of EA (W, X and Y). On the assumption that EC represent the same EA uni-modal distributions (though truncated by the haircut) as those observed, they were converted to putative EA (Fig. 2, right panel) by the algorithm described above (see Materials and methods). The resulting extrapolated EA (longer than the haircut) distributions were thereby extended to lengths of 77.5, 87.5 and 127.5 cm for samples W, X and Y, respectively.
Major-axis diameter changes along (within) and between EA
As previously shown for CA, 9 EA major-axis diameters similarly increase from their proximal clubs to a maximum near their distal ends, after which they sharply decrease. The maximal hair diameter can therefore be used as a boundary to distinguish the anagen tip, distally, from the hair shaft 'body', proximally. Fig. 1b and c illustrates how total (l), shaft-body (lm) and anagen tip (lt) lengths, and proximal (dp) and maximal (dm) diameters were defined for hair groups and how their rates of increases within and between fibres were calculated. Numbers of fibres and their measured data, for long and short (') hair groups are listed in Table 2. These are in fact just the two extreme hair-group lengths, out of several (numbers in parentheses next to donors) that were usually measured for each donor; these only are listed in the table to allow direct comparisons between all of the donors. Rates of shaft body diameter increases toward their maxima (1.03 ± 0.52 µm/cm and 1.20 ± 0.54 µm/cm, sw and sw', respectively) are linear in general (data not shown) and not significantly different. Their combined rate of increase is 1.11 ± 0.53 µm/cm, or 1.31 ± 0.72%/cm.
Estimates of hair diameter changes between fibres (Table 2, last two columns) used proximal and maximal diameters as natural reference points. Proximal diameter increases (sp) between short and long fibres average 0.84 ± 0.62 µm/cm (1.02 ± 0.76%/cm) and those of maximal (sm) diameters 1.44 ± 0.67 µm/cm (1.40 ± 0.66%/cm). These are significantly different from 0 (P < 0.0001, binomial distribution) and from each other (P < 0.001, Student's t-test, paired samples).
Regression analysis of proximal and maximal diameter changes between fibres of the same donors with six or more measured EA length classes (Table 2, numbers in parentheses for M, U, W and Y) are consistent with linear increases (Fig. 3). The correlations between maximal diameters and hair lengths are stronger than those for proximal diameters, consistent with the overall larger sm as opposed to sp values of Table 2. Diameter changes between fibres as determined by linear regression slopes (Fig. 3) are correlated with their respective sp and sm values (Table 2) for the same donors (r =0.97, P < 0.01, Spearman rank correlation). This implies that the sp and sm values, which were calculated using just the two extreme hair groups, are fairly robust estimates.
Anagen tip vs. EA length correlation
Anagen tip lengths (llm for each hair group in Table 2), were found to be linearly correlated (Fig. 4) with EA lengths (r = 0.73, P < 0.0001, n = 28). Their mean length, as derived from the slope of the regression, subsumes 22% of the mean EA lengths. A similar plot, using anagen tip lengths of all of the hair groups measured (n = 65, sum of numbers in parentheses in Table 2) yields a stronger correlation (r = 0.83), with the tips representing 27% of the whole fibre lengths.
Discussion Go to: Choose Top of page Introduction Materials and methods Results Discussion << References
In a previous work on hair diameters along their shafts, Hutchinson and Thompson, 9 measured major- and minor-axes diameters of EC and CA (called telogen and anagen hairs, respectively) and by collating regions of overlapping diameters were able to produce diameter profiles of presumptive EA (full-length hairs). These authors found proximally decreasing major-axis diameters, although to a lesser extent than those described here (0.45%/cm vs. 1.31%/cm). However, they did not recognize the existence of EA of different lengths, and consequently the correlations between EA lengths and their diameters and anagen tip lengths were not described. With hindsight, the results of this study show that the methodology of collating different EC and CA to produce composite EA profiles is likely to conceal length differences between EA and thus perpetuates a misconception that they are basically of the same length. A large EC and a small CA, for example, may have matching diameters after relatively long and short anagen phase durations, respectively, rather than after the same anagen durations as thought. 9 This will lead to a single composite EA length that is intermediate between their actual disparate ones. Moreover, as anagen durations can be arbitrarily redefined by shifting to locations of overlapping diameters, most combinations of similar EC and CA will converge to apparently uniform EA.
EC, which constitute 56% of shed hair and generally represent EA longer than the haircut were used here to extrapolate EA length distributions (Fig. 2). While the algorithm used is admittedly crude, it does seem adequate as a first approximation. On the one hand it does not make any assumptions with regards to the interpolated distributions, other than that they preserve the observed EA distributions, and on the other hand extrapolation of the EA histograms past their peaks by an exponential decay function seems qualitatively justified by the data, has a precedent (in ageing), 15 and at any rate approaches linearity for longer fibres, where it obtains practical expression. The maximal EA lengths that were obtained (77.5, 87.5 and 127.5 cm) for the three samples evaluated seem substantially longer than analogous values obtained by Hutchinson and Thompson (49 cm), which is understandable (see above). But they also suggest that EA are longer than generally accepted (2590 cm). Interestingly, several lines of investigation support an idea that hair is in fact substantially longer than currently believed. In a study of the EA length distribution of one older woman, without hair loss and who had never cut her hair, this author (unpublished) demonstrated a uni-modal EA length distribution extending to 125 cm. Similarly long EA, as determined by anagen durations of up to 11 years, were also found in one (and to date perhaps the only) well documented study, which followed individual (ageing) follicles over time. 8 Furthermore, a recent phenomenological study of women with long hair 16 concluded that hair lengths up to 183 cm fall within the linear range of a plot of hair lengths vs. numbers of observed women. In contrast, however, the strong correlation between mean EA and EC lengths as described here suggests that hair styles are not arbitrary, but that women 'perceive' the peak lengths of their EA distributions and cut their hair to approximately two to three times this length. Long anagen tips ( 25% of the total EA length as found here), would become increasingly more obvious in longer haircuts and may thus assist in that perception. Further investigations would be required to resolve these matters. Proximal diameter measurements of EA and (individually measured) EC could probably settle the problem of completely describing the length distribution of a hair mass despite the haircut.
The use of EA longer than 10 cm for diameter measurements effectively excluded putative miniaturized fibres, or small nonminiaturized fibres found at the hair mass periphery, from further consideration and thus the correlations between EA diameters and their length apply to terminal fibres. A similar correlation has been described previously (together with positively skewed EA length distributions described as log-normal) in male Androgenetic Alopecia and ageing 7,8, 15 but was attributed to miniaturization. It would be unreasonable to suppose that the correlation found here is also due to miniaturization for several reasons. In the first place, the broad distributions of lengths and diameters that were found would imply that if miniaturization was the responsible factor it would have started several cycles previously, which would be impossible, given the young age of the donors. 17 A supposition of early onset Androgenetic Alopecia would be likewise rejected, and further ruled out on the grounds that the prevalence of EA in shed hair (significant subpopulations in 22/25 donors) would make this the normal, rather than the rare condition it is currently thought to be. 3,18 Finally, as hair diameter normally increases to young adulthood, possibly even to the third decade, 1921 an assumption of 'normally occurring early onset' Androgenetic Alopecia would be incongruous, as it would imply that hair diameter and length are negatively correlated in young women, and was not found to be the case. Thus, it seems inescapable that EA of different lengths represent normal fibres under natural growth conditions.
Because the correlation between normal EA diameters and lengths would make such fibres indistinguishable from miniaturizing hair, it is tempting to speculate that Androgenetic Alopecia may represent a shift of the normal EA distribution toward shorter fibres. Such a shift would appear to move the focus in Androgenetic Alopecia from individually affected follicles to regions of the scalp or perhaps to the entire scalp. The patterned hair loss in male Androgenetic Alopecia appears to gratuitously support this idea, and it would include hair loss in ageing as a natural rather than pathological phenomenon, which seems preferable. It is, however, toward understanding female Androgenetic Alopecia that this concept could be most usefully applied at this time.
Although classical miniaturization was originally described in female Androgenetic Alopecia, where bimodal length and diameter distributions were observed, 2,3 recent studies 21,22 failed to demonstrate a (strong) correlation between hair density and diameter decreases in this condition. A rapid type of miniaturization has therefore been proposed to explain this apparent anomaly in female Androgenetic Alopecia. 17,21 In its most extreme form 17 rapid miniaturization is postulated to take place and be completed between catagen of one follicular cycle and (early) anagen I of the next. It further stipulates that hair diameter decreases (or any diameter changes for that matter) cannot occur during hair growth out of the scalp (anagen VI) on the grounds that this is determined by rapidly amplifying (matrix) cells derived only once per follicular cycle from stem cells at the bulge region. 23 The timing of the switch from normal to miniaturized follicles specifically dissociates observable hair growth from diameter changes and thus circumvents the problem of differentiating between miniaturizing and normal hair. However, no positive evidence has been produced to demonstrate rapid miniaturization so far, and the assumption that diameter changes do not occur during anagen VI (though possibly not critical for rapid-miniaturization, 21,22 ) is clearly contradicted by this, and other studies. 9,12, 13 Thus, it should be regarded as speculative at this time.
EA of differing lengths (the existence of which rapid miniaturization would paradoxically confirm as normal), could possibly explain the anomalous results found in female Androgenetic Alopecia without requiring rapid miniaturization. Slow miniaturization, operating on the entire scalp, could set up a cascade of EA lengths that might easily last over several follicular cycles. The mean diameters of the cascade, being correlated with fibre lengths as they are, would however, remain relatively steady. The shortest hair would eventually become lost (unobserved) and thus apparent hair density would decrease without concomitant diameter decreases. An exception to the cascade, which may be of low significance (and would remain mostly latent so long as the hair is cut), would be found with the largest fibres. Interestingly, preferential diameter decreases of large fibres in female Androgenetic Alopecia have been described. 21
EA, which have not been previously recognized as a class of hair
