INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

TIMELY ACTIVATION OF FETAL adrenal steroidogenesis plays a key role in triggering the maturation of other organ systems before birth (22). The development and the integration of the various components of the system that regulates adrenal function, the hypothalamic-pituitary-adrenal (HPA) axis, have been extensively studied and have come to be understood in fairly general terms (e.g., Refs. 11, 23, 27). Although many aspects of the activation of fetal adrenal steroidogenesis have been elucidated, such as the characteristic, progressive increase in fetal plasma glucocorticoids over the final days of gestation, other specific details have remained elusive. For example, there is still much to be learned about the ultimate input or signal that drives the cascade of events that results in the rapid escalation of glucocorticoids.

Many animal models that have been used to study development of the HPA axis, including ovine and primate models, have measured developmental changes in single fetuses. Although the study of single fetuses does reflect the majority of human pregnancies, it may overlook obscure important differences that occur in multiple pregnancies. Given the role of fetal steroids in preparing the fetus for extrauterine life, one important question regarding multiple pregnancies is whether the adrenals of the fetuses are activated synchronously. One aim of the present study was to measure concentrations of cortisol in the plasma of twin sheep fetuses over the final 2 wk of gestation to determine whether the adrenals are activated in synchrony and a coordinated manner or, if not, whether plasma glucocorticoids in one twin consistently exceed those of the other.

In the event that activation of the adrenals of twins takes place within different time frames, the study was further designed to exploit this difference as a means of providing insight into the potential mechanisms by which activation is effected. This was done by also measuring plasma concentrations of adrenocorticotropic hormone (ACTH) and estrogens; blood PO₂, PCO₂, and pH; and the intrinsic activity of the pituitary and adrenal cells in vitro at the end of the study.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Animals. All surgical and postsurgical procedures in these studies followed accepted veterinary medical practices and were approved by the Animal Care and Use Committee of this institution. Mixed-breed pregnant ewes with recorded dates of mating and bearing twins were used. In these sheep the length of gestation of singleton lambs typically runs 145 days. Surgery was performed on six ewes at 126 ± 1 days of gestation (Table 1). The sheep were anesthetized with ketamine and halothane and maintained under halothane anesthesia. With the use of strictly aseptic procedures, saline-filled polyvinyl catheters were implanted into the amniotic sac and into the descending aortas of both twins via both femoral arteries in each. After careful closure of the uterus, the catheters were led subcutaneously to the ewe's flank, where they were exteriorized. Catheters were also placed in one femoral artery and vein of the ewe and exteriorized at the same site as the fetal catheters. All catheters were protected by being wrapped in a sterile gauze pad, covered with a latex glove, and held in place by protective elastic mesh. Prophylactic intravenous antibiotics (1.5 mg/kg gentamicin and 20 mg/kg ampicillin) were administered at the time of surgery and once per day for 3 days afterward.

After recovery from anesthesia, the sheep were kept indoors in pens that allowed free movement. Food and water were available ad libitum. At no time did any of the sheep display any sign of illness or distress, and all procedures took place in the sheep pens.

Blood-sampling procedures. After at least 3 days of recovery (5 ± 1 days, mean ± SE), daily (between 0700 and 0900) sampling of fetal and maternal blood was commenced. Samples of 4 ml were drawn, and the volume was replaced by an equivalent amount of sterile isotonic saline. Partial pressures of O₂ and CO₂ in the blood and blood pH were measured immediately by automated blood gas analysis (Radiometer, Copenhagen, Denmark). Plasma cortisol concentration was also measured by radioimmunoassay (RIA) directly after sampling; remaining plasma was stored at 80°C until later analysis.

Time course of study. The study was initially designed to draw daily blood samples until as late in gestation as possible, but before the onset of labor, and then to recover the fetal pituitary and adrenals for in vitro studies. Based on previously published data (20, 28), the criterion that we established as being indicative of imminent labor was two successive measurements (i.e., 24 h) of plasma cortisol 50 ng/ml in either fetus, which experience has now taught us (see RESULTS) was not an appropriate indicator for twin sheep of this flock.

In vitro experiments. Fetuses, delivered by cesarean section, and neonates were killed by an overdose of intravenous pentobarbital sodium. The pituitary and adrenals were removed, dissociated, and cultured according to previously described methods (15, 34). Because there is evidence for an endogenous inhibitor of the adrenal responses to ACTH, the effects of which diminish with time in culture (14), and the function of pituitary cells is also altered by factors in the in vivo environment from which the tissue is obtained, we used cultured dissociated cells to examine inherent activity of the cells themselves. The left and right adrenals of each fetus were combined to yield a preparation of cells for each individual fetus. The pituitary cells of each individual were cultured separately. Briefly, the pituitary and adrenal glands were dissected, minced, and then placed in a digestion solution containing collagenase (type I for adrenal or type II for pituitary; both from Worthington, Freehold, NJ) and deoxyribonuclease (Sigma, St. Louis, MO). The cells were washed by centrifugation three times in culture medium (Dulbecco's modified Eagle's medium and Ham's F-12 medium mixed 1:1, to which fetal calf serum, penicillin, and streptomycin are added; all from GIBCO BRL, Grand Island, NY). The cells were plated in 48-well culture plates (Falcon, Franklin Lakes, NJ) in 1.0-ml culture medium (2 × 10⁵ pituitary cells/well; 1 × 10⁵ adrenal cells/well) and cultured for 3-4 days at 37°C and 5% CO₂. Cells were then washed three times in incubation medium [Dulbecco's modified Eagle's medium and Ham's F-12 medium mixed 1:1, to which Polypep (Sigma) is added to 0.2%], equilibrated to serum-free conditions for 1 h at 37°C and 5% CO₂, washed, and then incubated under test conditions. Duplicate wells of pituitary cells were incubated for 3 h in 250 µl (final total volume) incubation medium, containing either vehicle or maximum stimulating concentrations (100 nM) of ovine corticotropin-releasing hormone (CRH) or arginine vasopressin (AVP; Peninsula Laboratories, Belmont, CA). Lesser concentrations of CRH (10 nM) and AVP (1 and 10 nM) were also included in some experiments. Because of the limited number of cells obtained from individual pituitaries, only the highest concentrations of the peptides were tested in all experiments. Duplicate wells of adrenal cells were incubated for 6 h in 250 µl incubation medium containing either vehicle or human ACTH (10¹¹, 10¹⁰, 10⁹, and 10⁸ M; Peninsula). At the end of the incubation, media were removed and frozen at 80°C for later analysis. In each experiment, the pituitary cells that had been incubated with vehicle only (control cells) were solubilized in 0.01% Nonidet P-40 detergent, and the extract was frozen for later assay of cellular ACTH content.

Analyses. Plasma cortisol and cortisol in the incubation media of experiments with adrenal cells were measured by RIA (32). Cortisol in plasma samples was extracted into dichloromethane and dried before assay. When cortisol was assayed in samples of incubation medium, the samples were not extracted but the RIA standard curve was also made in incubation medium. Plasma samples were assayed for cortisol twice, once on the day they were drawn (to determine the timing of delivery) and a second time in one of three assays, in each of which was consolidated one-third of the total number of samples for simultaneous assay (to minimize variability). The results of the assays were consistent, and those of the large combined assays were used for statistical and graphical purposes.

ACTH in the incubation medium and extracts of the in vitro experiments were measured by RIA as previously described (10). Plasma ACTH was measured by a two-site immunoradiometric assay (IRMA) using monoclonal antibodies as previously described (35) and used to measure ACTH in fetal sheep plasma (9). The IRMA specifically measures ACTH-(139) and was used in these studies because this form of ACTH is the most potent in terms of agonist biological activity at the fetal sheep adrenal (31) and therefore the most relevant measurement for these studies.

Estrogens in plasma were measured with a commercially available RIA kit (ICN, Costa Mesa, CA) after having been extracted into ethyl acetate:hexane, 3:2, and dried. The antiserum used in this assay recognizes both 17-estradiol and estrone. The assay was performed as specified by the manufacturer, with a sensitivity of 2.5 pg/tube. All samples of plasma undergoing assay for estrogens were assayed in the same RIA.

All data are presented as means ± SE. Duplicate measurements from the in vitro experiments were averaged in all cases to yield an n of 1 for each experiment. Organ weights and body measures were compared between twins by paired t-test. Correlations between plasma cortisol concentrations and factors that might influence its secretion were analyzed by a Pearson test of correlation, followed by assessment of probabilities, adjusted by the Bonferroni method. All other results were analyzed by two-way analysis of variance and, where significant effects were indicated, specific differences were assessed by Tukey's post hoc test. Differences were considered significant for P < 0.05. In addition, analysis of the concentration-response relationship for ACTH on cortisol secretion by adrenal cells in culture was continued by nonlinear regression analysis of the curves. This was accomplished by fitting the curves to a sigmoid equation and determining the best fit by repeated iteration. The curves for adrenal responses in these cells were compared between cells obtained from group A fetuses and group B fetuses (see below) in terms of mean effective concentrations (EC₅₀) and maximum ACTH-secretory responses. Differences were considered significant if 95% confidence intervals did not overlap.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

All 6 ewes and 12 fetuses survived the procedures of this study until the prescribed euthanasia. Maternal signs, such as appetite, indicated that the ewes remained healthy during the study.

Fetal well-being, as assessed by measurement of blood gases, was also maintained throughout the study. Partial pressures of O₂ and CO₂ in fetal blood averaged 18.9 ± 5.1 and 50.7 ± 2.7 mmHg (mean ± SD), respectively. Arterial pH averaged 7.32 ± 0.05. In individual animals there were occasional occurrences of low O₂ (<15 mmHg), high CO₂ (>53 mmHg), or low pH (<7.29), but in no case were there simultaneous measurements of abnormal O₂, CO₂, and pH indicating anything pathological.

Table 1 summarizes the gestational ages of the twins at various stages of the study. Because it was necessary to have a reference time point to compare data in all animals, day 0 was arbitrarily defined as the first day plasma cortisol was 50 ng/ml in either fetus (Table 1). Elective delivery by cesarean section was performed, as originally planned, when plasma cortisol was >50 ng/ml for two successive daily measurements in only two sets of twins (sets 1 and 2). In twins (set 3), the fetuses were removed by cesarean section on the first day plasma cortisol was 50 ng/ml because a later amniotic pressure reading indicated the ewe was in labor. In twins set 4, plasma cortisol went above 50 ng/ml in both twins, but for 1 day only before spontaneous birth, and there were no outward signs of labor on the evening before birth. Set 5 delivered spontaneously before plasma cortisol, based on the daily assay, exceeded 50 ng/ml in two successive measurements. In set 6, plasma cortisol was >50 ng/ml on the second day before birth but <50 ng/ml on the day before birth and >50 ng/ml again on the morning of birth.

Plasma cortisol concentrations. Cortisol concentrations increased steadily with advancing gestational age. For purposes of data analysis we defined baseline fetal plasma cortisol concentration as the mean of all fetal concentrations where fetal plasma cortisol concentration was less than the simultaneous maternal cortisol concentration (=7.8 ng/ml). We defined the threshold of increased fetal plasma concentration as being baseline + 2 SD (=13.4 ng/ml). In each set of twins, the fetus in which plasma cortisol exceeded threshold first was designated as fetus A.

One important question addressed by these studies is whether activation of adrenal steroidogenesis occurs synchronously in twins and, if not, whether it occurs in parallel, but offset by a constant time quotient. Basically, we asked whether, once threshold was passed by one twin, that twin continued to produce higher plasma concentrations of cortisol than its sibling. When the simultaneous plasma concentrations of the twins are separated solely on the basis of which twin first crossed threshold, it becomes clear that adrenocortical activity of the twin that crossed first takes a lead over that of its sibling (Fig. 1). Considering the data from simultaneously drawn blood samples in all the pairs of twins, we found that once either twin crossed threshold, its plasma cortisol concentration was higher than the simultaneous concentration in the other twin 81% of the time (29 of 36 pairs of subsequent simultaneous measurements). In addition, analysis of variance of the average plasma cortisol concentrations in the two groups of twins, A and B, indicates significant effects of day (essentially gestational age) and group on plasma cortisol and a significant interaction between day and group (Fig. 1). In all sets of twins, plasma cortisol concentrations in group A fetuses reached 50 ng/ml before (5 of 6) or on the same day (1 of 6) as those in group B fetuses. Interestingly, it was also noted that plasma cortisol in group B twins often never attained the levels in their siblings. In two of three sets of twins that delivered live lambs, plasma cortisol levels in the B twin remained remarkably low. In one, plasma cortisol crossed 50 ng/ml (56.5) only within 24 h of birth, with the previously highest cortisol having been 37.8 ng/ml. In the other B twin, cortisol was only 45.3 ng/ml on the morning of birth (which occurred 3 h after the sample was drawn), with the previously highest cortisol having been 11.5 ng/ml and plasma cortisol having actually decreased (6.4 ng/ml) on the day before birth.

View larger version (30K): [in this window] [in a new window]   Fig. 1.   Plasma cortisol concentrations as a function of gestational age, with the twins grouped according to which first crossed the threshold for increased cortisol (group A fetuses; ). Solid horizontal line at 7.8 ng/ml represents baseline plasma cortisol concentration, defined as the average of all fetal plasma cortisol concentrations where fetal plasma concentration was less than simultaneous measurement of maternal cortisol. Hatched area represents ± 2 SD of those measurements. Threshold for increased cortisol was defined as the first measurement of cortisol greater than baseline + 2 SD (=13.4 ng/ml). In each pair of twins, the twin whose cortisol was first over this mark was considered fetus A and the other, by default, fetus B (). Curved lines represent third-order regression analysis of the data. Symbols represent means ± SE (maternal concentrations indicated by triangles). +,* P < 0.05 for effect of time and difference between group A and B fetuses, respectively.

There was no correlation between plasma cortisol concentration and either blood O₂, CO₂, or pH (reckoned as such or as H⁺ concentration).

Plasma ACTH-(139) concentrations. Daily changes and differences between twins in the concentration of plasma ACTH-(139) did not present as clear an outcome as did plasma cortisol. Over the course of the entire experiment (i.e., simply comparing the concentrations of ACTH in the plasma of each fetus at the first sample drawn with that in the last drawn before labor), there was a significant increase in plasma ACTH-(139) (P < 0.01). After defining the threshold, indicating stimulated plasma ACTH in a manner similar to that for plasma cortisol [mean + 2 SD of all fetal ACTH-(139) measurements where fetal < maternal ACTH-(139) concentration], we found that all the fetuses that crossed the cortisol threshold before their sibling (group A) also crossed the ACTH threshold first. Group A fetuses did not cross the two thresholds on the same day, but five of six fetuses in group A crossed the cortisol threshold before they crossed the ACTH threshold (Table 1).

days 6

2

View larger version (31K): [in this window] [in a new window]   Fig. 2.   Plasma concentrations of ACTH-(139) as a function of gestational age, with the twins grouped according to which first crossed the threshold for increased cortisol (group A fetuses; ). In each set of twins, the sibling that crossed the cortisol threshold also first crossed the corresponding ACTH threshold. Circles represent the other sibling twin (group B fetuses); triangles represent simultaneous measurements in the ewe. Values are ±SE. Line at 8.85 pg/ml represents baseline ACTH secretion, and hatched area represents ± 2 SD, calculated as for cortisol in Fig. 1.

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Plasma cortisol concentrations as a function of gestational age, with the twins grouped according to sex and day 0 defined as the day on which the last sample was drawn before the ewe went into labor. In each of 4 sets of twins there was 1 male and 1 female fetus. Values are means ± SE. +,* P < 0.05 for effect of time and difference between sexes, respectively.

View this table: [in this window] [in a new window]   Table 2.   Plasma ACTH-(139) and total estrogens (17-estradiol plus estrone) as a function of gestational age, examining differences between sexes in male-female sets of twins and aligning data on day 0 as being the sample drawn before the ewe went into labor or on the day before spontaneous delivery

View larger version (18K): [in this window] [in a new window]   Fig. 4.   Comparisons of ACTH- and cortisol-secretory responses in vitro between cells obtained from groups A and B fetuses. Values are means ± SE. Top: ACTH responses are by pituitary cells in response to maximum stimulation with corticotropin-releasing hormone (CRH; 100 nM) or AVP (100 nM) and are expressed as pg ACTH secreted per ng ACTH content. Bottom: cortisol responses are by adrenal cells in response to indicated concentration of ACTH and are expressed as ng secreted per well (100,000 cells). #,* P < 0.05 for effect of concentration of ACTH and difference between group A and B fetuses, respectively.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The immediate object of the present study was to determine whether concentrations of plasma cortisol increase at the same time and to the same extent in twin sheep fetuses as they approach term. Given the role in organ development played by systemic glucocorticoids (3, 22), this question relates to the overall preparation of fetuses for extrauterine life, as well as the timing of the specific endocrine event. The data show that activation of the fetal HPA axis, as indicated by the initial increase in cortisol over basal levels, does not occur simultaneously in twins. In addition, there was a significant separation between the twins in the timing of the overall increase, because plasma cortisol concentrations in the twins that crossed the threshold first were consistently higher than simultaneously measured concentrations in their siblings over the course of this study.

Given that each set of twins is of similar genetic composition and subject to nearly identical maternal and environmental factors in utero, these findings imply that in sheep the ultimate trigger and mechanisms for the onset of the increase in fetal plasma glucocorticoids resides in the fetus and not the mother or placenta. Kitts et al. (20) performed a study with mixed-breed twin pregnancies achieved by embryo transfer. They observed that activation of C-21 adrenal steroidogenesis occurred first and remained ahead in the sibling of the breed with the shorter period of gestation. In the earlier study, the timing of steroidogenic activation was thus attributed to the genetic composition of the fetuses; it is noteworthy that a similar phenomenon was observed in the present study despite the greater genetic similarity of the twins.

The differences in plasma cortisol between the twins also suggest that in multiple pregnancies there is little communication between the fetuses in this regard. Even before labor, differences of >20 ng/ml between twin fetuses were not uncommon, and in one set of twins there were 2 days when the difference was >50 ng/ml. These data, with naturally occurring changes in plasma cortisol, resemble those of a study by Brooks et al. (8) in which ACTH was infused into one fetus in twin pregnancies. In that study, plasma cortisol increased only in the infused fetus. Similarly, Kitts et al. (20) measured significant differences in cortisol between fetal twins in their embryo-transfer experiments. The widely divergent values for plasma cortisol obtained at simultaneous samplings from maternal and both fetal arterial circulations is consistent with the observations of Beitins et al. (6) in their direct studies of resistance to diffusion by the ovine placenta. They concluded that negligible amounts of cortisol cross from the maternal to the fetal circulation and that the small amount that crosses in the opposite direction contributes very little to the maternal cortisol concentration.

If the trigger for the stimulation of adrenal steroidogenic activity is resident in the fetus, then analysis of the differences between twins may yield clues to the nature of the trigger. In the present study this was noted in two ways, in terms of the anatomic level at which the timing mechanism may reside and in terms of the mechanisms that might operate to implement or act as a result of the trigger.

The responses of cells cultured from the individual twins (fetuses and neonates) postmortem shed light on the level at which key mechanisms for the timing of the increase in cortisol may occur. The difference between the cortisol responses to ACTH of the adrenal cells of the two groups suggests that inherent differences in the development of fetal adrenals, rendering them more responsive to ACTH, contribute more to the increase in plasma cortisol than do changes in the responsiveness of pituitary cells to hypothalamic factors. Studies of cortisol responses to exogenous ACTH in vivo indicate that adrenals become more responsive with age (33), and it is not unreasonable to infer that with twins it is possible for the onset of the increase in responsiveness to occur in the adrenals of one fetus earlier than the other. The finding that group A adrenal cells had a greater cortisol response to ACTH than did those of group B also supports the concept that the elevated plasma cortisol measurements in group A were more likely due to increased secretion by the adrenals rather than decreased clearance of cortisol, although these results by themselves are not conclusive.

Interestingly, in all four sets of twins with male and female fetuses, the males crossed the cortisol and ACTH thresholds first and had significantly higher plasma cortisol concentrations (group A) than the female siblings (all group B). This apparent sex difference suggested a potential mechanism for the timing of the increase in cortisol involving sex steroids. The absence of a difference between twins in concentrations of total estrogens argues against a role for total estrogens. The present findings are similar to more extensive results reported by Kitts et al. (19, 20), in which there was a difference in plasma cortisol concentrations between the Rambouillet and Finnish Landrace siblings but no difference in unconjugated estrone, estradiol, androstenedione, or estrone sulfate and in which plasma concentrations of the estrogens also increased after those of cortisol.

The precipitous increase in plasma cortisol measured in the twins in the present study is consistent with previous measurements made in singleton fetuses approaching term (5). In twins, the differences over time and between groups were significant whether the timing of the crossing of plasma cortisol threshold or sex was the determining factor for grouping the siblings. Similarly, the differences were significant whether the day of the first plasma value of 50 ng cortisol/ml or the day before labor was used as day 0. These observations support the idea that the increases in cortisol are indeed different between the twins.

The measurements of plasma ACTH are somewhat more difficult to interpret. Although there was an overall correlation between simultaneous plasma ACTH-(139) and cortisol values, in contrast to the case with cortisol, there was no significant increase in plasma daily ACTH over the course of the study nor any significant difference between twins. The absence of any effect of time (i.e., gestational age) on plasma ACTH in the presence of an effect of time on cortisol may reflect increasing responsiveness with time of the adrenals to stimulation by ACTH. Thus ACTH could be increasing on a daily basis, with changes too small to be detected, at the same time that adrenal responses to ACTH are also increasing, the net effect being increases in cortisol of greater magnitude than those of ACTH and thereby significant. Another factor to be considered is that the pulsatility of ACTH secretion (2, 18) may have increased the variability of measurements and obscured daily changes. A third factor to be considered is the effect of steadily increasing cortisol on ACTH secretion. Negative feedback by glucocorticoids, even if attenuated at this stage of gestation, might provide enough inhibition of ACTH secretion to render daily changes in ACTH indistinguishable. With regard to this last consideration, it may be worth contrasting Figs. 1 and 2 over days 1 to +1, during which time plasma cortisol is clearly still increasing while ACTH concentrations are not.

Regarding the absence of differences in plasma ACTH concentrations between twins, the mechanisms described in the previous paragraph also may be responsible. In particular, the differences in sensitivity between the adrenal cells to ACTH of groups A and B may be sufficient to explain why there can be a difference on plasma cortisol concentrations without a simultaneous significant difference in plasma ACTH.

One unexpected finding was the extent to which plasma cortisol remained low in some of the individual twins in group B. This was true even among twins that were allowed to continue to birth, in all cases of which the neonates were healthy, breathing, and nursing. The precise role of plasma cortisol for maturation of various organ systems in terms of concentrations and duration of exposure is still a matter of study. Prolonged survival studies of the live-born lambs to test whether all organ systems were fully developed were not performed in the present study. Research suggests that for maturation of the lungs, at least, glucocorticoids, as a single component, may be more critical at earlier deliveries than at term (3). That would appear to be the case in the present studies.

Another interesting finding was the absence of correlation between fetal blood gas parameters and plasma concentrations of cortisol. Among the known stimuli in adult and fetal animals of HPA axis activity are hypoxia, hypercapnia, and acidemia (e.g., Refs. 24, 29, 30, 36). None of these conditions by themselves were associated with increased plasma cortisol in the present study. The elevations of H⁺ concentration and CO₂ levels were modest. As such, the results are consistent with those of Chen and Wood (13), who reported that fetal sheep (123-127 days gestational age), whose CO₂ tensions were elevated to 55.2 and pH levels were decreased to 7.257 as a result of increasing the fraction of CO₂ in the maternal inspired air, maintained normal plasma ACTH and cortisol concentrations. Hypoxemia is a more common cause of increased pituitary-adrenal activity in fetuses, and several experimental models have been employed to study the effects of decreased oxygen (1, 7, 12, 17). Over a period of up to 24 h, there is a clear and sustained increase in cortisol in response to hypoxia (1, 7, 12, 17). In contrast, in a model of chronic fetal hypoxia, Harvey and co-workers (16) found no effect on plasma ACTH or cortisol concentrations. This would suggest the existence of dissociable response mechanisms to hypoxia in fetuses, which may operate in separate time domains. The results of the present study would be entirely consistent with this description. Interestingly, the one fetus with the lowest average measurements of oxygen also had the lowest average measurements of cortisol, with both measurements being consistent over an entire period of 5 days of measurements. The responses in this animal would appear to fit a model of chronic hypoxia as in the experiments of Harvey et al. (16).

Perspectives