In mammals, the circadian timing system has proven to be involved in the daily variation of almost any physiological and psychological variable evaluated so far (Weitzman et al. 1978; Wirz-Justice 1987; Johnson et al. 1992; Dijk and Czeisler 1995; Boivin et al. 1997). The rhythm of the circadian pacemaker comprised in the nuclei suprachiasmatici (SCN) becomes entrained to the environmental 24-hour rhythm, predominantly by two zeitgebers, light and melatonin (Dawson and Armstrong 1996).

Melatonin influences the circadian phase in humans inversely compared to light. In particular, this drug delays circadian rhythms when administered in the morning and advances them when administered in the afternoon or early evening according to a phase response curve (PRC), which is nearly opposite in phase to the PRC's for light exposure (Lewy et al. 1992). Melatonin has proven to exert a chronobiotic mode of action (Dawson and Armstrong 1996) by its ability to facilitate the post-flight adaptation to jet-lag (Arendt et al. 1986), to phase advance the sleep of patients suffering from a phase-delay syndrome (Dahlitz et al. 1991; Tzischinsky et al. 1993), and to re-entrain the sleep-wake cycle of patients with a non-24-hour rhythm, such as in the blind, to the environmental light-dark cycle (Arendt et al. 1988; Lapierre and Dumont 1995; McArthur et al. 1996).

However, in only one of the clinical studies (Haimov et al. 1995), in which exogenous melatonin proved usefulness, did low endogenous melatonin excretion predict response (Garfinkel et al. 1995; Hughes et al. 1998; Lushington et al. 1998; Youngstedt et al. 1998). Furthermore, normative data as to the amount of melatonin secretion demonstrate a huge inter-individual variability in humans (Dawson et al. 1992; Bergiannaki et al. 1995; Smith et al. 1997). Thus, existing normative data on melatonin excretion in a “healthy population” might only be considered normal in the sense that they are representative for the normal scatter. But since there is yet no pathology that can be attributed to either high or low melatonin levels, “normative” in this context may not necessarily indicate being healthy. The question arises, does a melatonin deficit or a hypopinealism exist at all?

In a previously reported pilot-study we showed that the degree of pineal calcification (DOC) correlated positively to the incidence of chronic daytime tiredness as well as the subjective perception of sleep disturbances (Kunz et al. 1998). The basic assumption in that study was the hypothesis that an increasing degree of pineal calcification, such as with progressing age, would consecutively lead to decreasing melatonin secretion and thereby contribute to an instability of the circadian timing system, including circadian driven components of the sleep-wake cycle, such as circadian sleep propensity and continuity (Dijk and Cajochen 1997; Dijk and Czeisler 1995).

Even though this hypothesis seems to be plausible, it is not in agreement with the literature. It is generally stated that no correlation can be demonstrated between the occurrence of pineal calcification and a reduced melatonin production (Arendt 1995). But, the two studies frequently cited in that context used predominantly skull X-ray, which provides only little information on the quantity of calcification-density and no information about the size of the remaining non-calcified tissue (Bojkowski and Arendt 1990; Commentz et al. 1986).

The adult shape and size of the pineal gland are, presumably genetically, determined early in life and reach their final stage around the first year of life (Wetterberg et al. 1983; Schmidt et al. 1995). The weight and dimensions are highly variable—up to a 20-fold inter-individual variance (Vollrath 1981; Hasegawa et al. 1987; Bhatnagar 1990; Schmidt et al. 1995). Therefore, the notion of an “average-size” is not adequate. Since most of the pineal tissue consists of melatonin producing pinealocytes (Reiter 1981), the high inter-individual variability as to the amount of melatonin secretion may easily be explained by the variability of the amount of (uncalcified) pineal tissue. In addition, since the size of the pineal calcification does not give any information on the remaining active pineal tissue, it is not surprising that the size of pineal calcification itself is not negatively associated to melatonin excretion.

The presented study was aimed at establishing a relationship between melatonin secretion and the size of uncalcified pineal volume.


Study Population

The study population of 26 individuals consisted of 23, otherwise healthy patients, who had contacted our outpatient clinic for sleep-related complaints, and three healthy subjects without sleep-related problems (16 female, 10 male; mean-age 44 years, SD: 15; range 17–72 years). Diagnoses according to ICSD-criteria (International Classification of Sleep Disorders) were: healthy (n = 3), sleep state misperception (n = 4), idiopathic insomnia (n = 7), psychophysiological insomnia (n = 4), and restless legs syndrome and/or periodic limb movement disorder (RLS/PLMD) (n = 8). Whereas 23 subjects were completely free of medication, three subjects had been taking thyroxin for years because of hypothyroidism.

Extensive examinations had to show no evidence of a neurologic, psychiatric or other somatic disease in any of the included subjects. There had to be no evidence of drug or alcohol abuse. All participants were instructed to abstain from alcohol consumption and over-the-counter medications during all study procedures. Computer Tomography (CT)—scans in all subjects had to be considered as revealing no pathological findings. Performed polysomnograpy (PSG) had to rule out any other diagnosis except for primary insomnia or RLS and/or PLMD. Patients, in whom diagnosis was RLS/PLMD, underwent specific evaluation. Patients were only included when the disorder was considered to be idiopathic.

Special care was taken that subjects did not take any substances which are known to interfere with melatonin production and/or secretion, but which, among the general population, are sometimes not considered to be “medication,” e.g., with respect to sleep: benzodiazepines (McIntyre et al. 1988), antidepressants (Skene et al. 1994), beta-blockers (Cowen et al. 1985), and with respect to pain: anti-inflammatory drugs (Surrall et al. 1987; Murphy et al. 1994). Drugscreening in urine included benzodiazepines, barbiturates, cannabinoids, amphetamines, cocain and opiates.

The subjects provided their written informed consent. The experimental protocol was approved by the Ethics Committee of the Freie Universität Berlin.

Computed Tomography and Image Analysis

In order to estimate pineal volume and DOC, cranial CTs were performed on a Siemens Somatom DRG or DRH in all subjects with the pineal gland covered by 4-millimeter thick, adjacent slices. The evaluation of the imaging data was performed in duplicate in a blinded fashion. The pineal gland was outlined using the image analysis function provided by the CT-machine, thus providing the area of the organ in square centimeters. This analysis was performed on all slices on which the organ, or parts of it, were displayed. In the case that the pineal gland was represented in more than one slice, the area values of the individual region of interest (ROI) were added to yield the total organ area.

An automatic image analysis of this ROI provided the maximum attenuation, or density, in Hounsfield Units (HU). This maximum organ density was grouped on a 5-category scale: 0–50 HU = 0; 51–150 HU = 1; 151–250 HU = 2; 251–350 HU = 3; 351–1000 HU = 4. Within the gland, parts were defined as calcified and uncalcified by visual inspection. Thus, the defined calcified portion was related to the whole organ area, and its relative portion estimated on a 4-category scale: 0–24% = 0; 25–49% = 1; 50–74% = 2; 75–100% = 3.

The imaging data was combined to a score of calcified pineal volume by adding the scale values of maximum organ density and the scale-values of calcified organ portion. This resulted in a scale from 0 to 7. By dividing the scale-value of calcification by 7 (scale maximum), it is possible to obtain the degree of calcification (DOC). This definition of DOC is a slightly modified version of the previously introduced method (Kunz et al. 1998). Here, DOC 0/7 represents an organ with no calcification, and DOC 7/7 a completely calcified pineal gland of high density. In this respect, DOC represents an estimation of the relation between calcified and total size of pineal tissue. For example: maximal density: 285 HU, calcified portion, 60%: 3 + 2 = DOC 5.

In order to estimate the calcified and uncalcified pineal volume, the DOC was then related to the calculated area of the pineal in two ways: firstly, by multiplication of DOC with the organ area in order to estimate the calcified, presumably inactive pineal tissue (e.g., area: 0.75cm2, DOC 3/7: 0.75 × 3/7 = 0.321 calcified tissue); secondly, 1 − DOC was multiplied by the organ area in order to estimate the uncalcified, presumably active pineal (e.g., area: 0.75cm2, DOC 3/7: 0.75 × (1 − 3/7) = 0.429 uncalcified tissue). Since, in this respect, calcified and uncalcified pineal tissue are partly based on estimated values, they have arbitrary volume units (VU).

Melatonin Measurements

Besides very few pathological exceptions, in which intestinal enterochromaffin cells contribute to serum melatonin, the only relevant secretion site for melatonin in the human body is the pineal gland (Tetsuo et al. 1982; Arendt 1995). About 80% of the pineal gland consists of pinealocytes with the main secretory product being melatonin (Karasek 1983; Reiter 1981). About 90% of secreted melatonin are excreted in urine as 6-sulfatoxymelatonin (aMT6s) with a plasma half-life of 10–45 minutes (Arendt 1995). Thus, measuring a 24-h amount of aMT6s in urine seems to be an appropriate method for a quantitative estimation of the secretory activity of the pineal to produce melatonin (Graham et al. 1998).

The excretion of aMT6s in urine was measured over a 32-hour period under controlled conditions while subjects were stationed on the ward for polysomnographic recording. Subjects were asked to collect urine in five fractions over five consecutive time intervals: 1) 11.00 PM till 07.00 AM; 2) 07.00 AM till 11.00 AM; 3) 11.00 AM till 06.00 PM; 4) 06.00 PM till 11.00 PM; and 5) 11.00 PM till 07.00 AM.

To control for night-to-night variability of aMT6s, two night fractions—during which the largest quantity is excreted—were averaged. In line with literature, night-to-night intra-individual variability was low, not exceeding 15% in any of the subjects (e.g., Bergiannaki et al. 1995). A 24-h amount of aMT6s was obtained by adding the averaged night-time fraction to the three day-time fractions. All samples were analyzed in duplicate by IBL (Hamburg), using a highly sensitive, competitive ELISA kit (Pfluger and Minder 1996).

Statistical Analysis

Data were analyzed using Spearman's rho in order to compute the association between aMT6s and the variables age, sex, DOC, uncalcified, calcified, and the total size of the pineal gland. Applying linear regression analysis, the relationship between DOC, the total size, and the approximated uncalcified respectively to the calcified size of the pineal gland to melatonin excretion were quantified. Here, the 24-h amount of aMT6s was the dependent variable. Modelling strategy proceeded by including sex and age simultaneously with either DOC, the uncalcified, calcified and the total size of the pineal gland in the model.


Descriptive statistics are shown in Table 1 . In line with the literature, there was a high inter-individual variability in the size of the organ as well as the total amount of aMT6s excretion. Again, in line with the literature, there was no association between pineal size and age (R (correlation coefficient): −0.096; p = .321).

Table 1 Descriptives

The total size of the pineal gland was significantly and positively associated to aMT6s using Spearman's rho (R: 0.493; p = .005). However, the positive association between uncalcified pineal tissue and aMT6s yielded the highest correlation among all investigated parameters (R: 0.672; p < .001).

There was a significant and negative association between aMT6s and age (R: −0.504; p = .004), as well as a strong tendency towards a negative association between uncalcified pineal tissue and age, though not significant (R: −0.309; p = .062).

Using linear regression models, neither DOC nor the size of calcified pineal tissue were associated with the 24-hour amount of aMT6s in urine (Figure 1). In contrast, the model including age, sex, and the uncalcified pineal tissue (adjR2 (adjusted R square): 0.402; F: 6.601; p = .002) yielded significance (Figure 2).

Figure 1
figure 1

Melatonin excretion vs. calcified pineal. VU: volume units

Figure 2
figure 2

Melatonin excretion vs. uncalcified pineal. Fit line represents regression: dependent aMT6s in ng/24h; predictor: uncalcified pineal; VU: volume units

By determining betacoefficients in the model including sex, age, and total pineal size as predictors for aMT6s, age [b (standard betacoefficient): −0.470; p = .032], and pineal size (b: 0.428; p = .025) both yielded significance. Exchanging uncalcified pineal tissue for total pineal size within the model, uncalcified pineal tissue (b: 0.519; p = .005) was highly significant, but age was not (b: −0.278; p = .164).


The presented data confirm various earlier studies. Firstly, the size of pineal calcification is not associated with melatonin secretion (Bojkowski and Arendt 1990; Commentz et al. 1986). This is in line with the lack of association between the size of pineal calcification and pathologies. Secondly, there is no association between pineal size and age (Hasegawa et al. 1987), which confirms the predominant, genetical determination of pineal size (Schmidt et al. 1995; Bhatnagar 1990). Thirdly, there is a negative correlation between age and aMT6s (Sack et al. 1986; Sharma et al. 1989; Humbert and Pevet 1994; Zhdanova et al. 1997; Waldhauser et al. 1988; Myers and Badia 1995).

It was shown for the first time in the present study that an approximation of the size of uncalcified pineal tissue, presumably representing active pinealocytes, is significantly and positively associated with the total amount of 24-hour aMT6s excretion in urine. Most interestingly, the well-known negative correlation between age and aMT6s did not persist when data were adjusted to uncalcified pineal tissue. Thus, it is safe to conclude that the decrease of aMT6s excretion with age is predominantly due to increasing pineal calcification and not determined by the aging process per se.

To date, in spite of the structural and histochemical appearance of pineal calcifications being well described, little is known about their biogenesis (Alcolado et al. 1986). Schmid (1993) summarized evidence, further confirmed by Humbert and Pevet (1995), that calcification of pinealocytes results from death or degeneration of the cell itself, thus leading to an overall decrease in pineal activity. Besides, it has been shown that, with age, there is a significant reduction in both melatonin content and the total number of pinealocytes due to a reduction of light pinealocytes, which are known to be functional (Skene et al. 1990; Humbert and Pevet 1995). Occurring simultaneously, there is an increase of dark pinealocytes, which are characterized by an intranuclear deposition of calcium and with many signs of degeneration (Schmid et al. 1994; Humbert and Pevet 1995). With the present data, we may be able to add a link missing so far in the chain of arguments that an increased calcification of the pineal gland represents a decrease in the number of functioning pinealocytes, which results in a decreased ability of the pineal gland to produce melatonin.

What is the clinical impact of our finding? There is overwhelming evidence that melatonin exerts its chronobiotic effects via the SCN (Cassone et al. 1986). On the other hand, there are many hints pointing to the assumption that the SCN's sensitivity for light and melatonin varies with time (Thompson et al. 1990; Owen and Arendt 1992; Dubocovich et al. 1996) as well as under the influence of various modulators (Medanic and Gillette 1992; Rea et al. 1994; Duncan et al. 1998). It also varies between states (McIntyre et al. 1990), but moreover between individuals (Lewy et al. 1985; McArthur et al. 1996). It has even been shown that melatonin itself influences the sensitivity of its SCN receptors (McArthur et al. 1997).

Given that there is a tremendous, genetically determined, inter-individual variability as to the amount of melatonin excretion, the question arises whether there is an individual threshold for the amount of melatonin required. Moreover, it remains unexplored as to what degree the circadian timing system may adapt to decreased melatonin production as the result of increasing pineal calcification. As long as these questions remain unanswered the notion of a “melatonin deficit” lacks a firm conceptual basis. Theoretically, melatonin may not even be necessary at all to the functioning of the circadian timing system, as long as a person stays entrained by well-timed light exposure and/or sufficient strong social zeitgebers. But pinealectomy has proven to increase the SCN's sensitivity to phase-shifting properties of light (Cassone 1992), thereby suggesting that intra-individually decreased melatonin, as indicated by an increased pineal calcification, may be a vulnerability marker with respect to the stability of the circadian timing system.

The authors are unaware of any study on melatonin production that controlled for either the total size of the pineal gland or the size of uncalcified pineal volume. But, since there is a 20-fold variability in the size of the pineal gland as well as in melatonin production, normative data on melatonin excretion may be useless if the starting point of the intra-individual capability to produce melatonin, as indicated by the total size of the pineal gland, is unknown. We suggest that pineal DOC might prove to be a better marker of an intra-individual pineal deficit, with respect to melatonin secretion than the amount of melatonin secretion itself. It needs to be established whether this deficit marker is associated with pathologies, especially with respect to the circadian timing system in man (Czeisler et al. 1992; Myers and Badia 1995).

Of course, the presented concept is speculative, based on many assumptions, which need confirmation. But, our hypothetical framework might explain why low melatonin secretion does not predict response to melatonin replacement therapy (Garfinkel et al. 1995; Hughes et al. 1998; Lushington et al. 1998; Youngstedt et al. 1998). Some evidence for the validity of our hypothesis comes from the preliminary finding that high pineal DOC is related to disturbances of the sleep-wake cycle (Kunz et al. 1998).

In conclusion, the increasing degree of pineal calcification might be established in the future as an indicator of the intra-individual decrease of the pineal gland's melatonin production. If it will be confirmed that pineal DOC is a vulnerability-marker of disturbances of the sleep-wake cycle and/or the circadian timing system, pineal DOC might also prove to be an indicator of the outcome of melatonin replacement therapy.