A “Tale of Two Cities” – new version, neuroscience – style.

Peer review: “Katrin Amunts, Vadim Istomin, Axel Schleicher, Karl Zilles. Postnatal development of the human primary motor cortex: a quantitative cytoarchitectonic analysis” Anat. Embryol., (1995) 192:557-571.  and “K. Amunts, A. Schleicher, K.Zilles. Persistence of layer 4 in the primary motor cortex (area 4) of Children which Cerebral Palsy” (J. Brain Res. (1997), 38, 2, 247-260)

I. Why is this sudden attention to some 20+ years old papers?

Usually, finding a new reference to your own paper, especially if it was published a while ago, makes you feel good. After all, it is always pleasant to know that your small contribution to some area of science did not vanish without traces. But, unfortunately, such discovery does not always creates a happy feeling. That is the case when the referenced paper is used in a completely wrong context, which has nothing to do with results actually published. It is bad enough when not very scrupulous authors clearly demonstrate by their writing that they include reference to a paper without reading it. But it is much more unfortunate when people who use the reference for a wrong reason are former co-authors, namely K. Amunts, A. Schleicher and K. Zilles.

Let’s look at some examples. In [1], (p. 175, left column, last paragraph),  the reference to our paper implies that this technology is just another version of GLI “used in parcellation studies based on improved images of the laminar pattern”. (If you do not know what GLI is – please, read the review of GLI here).

Another reference in [2],  (p. 33, left column, last paragraph) states the same, but more directly: “GLI profiles were used… to analyze postnatal development of human primary motor cortex (Amunts et all, 1995)”. Unfortunately, the referenced paper has nothing to do with GLI.

Another example: in [3], (p. 14, the end of the first paragraph), we read: “The establishment of the “grey level index (GLI)” as a reliable estimate of cell body packing density (Wree, Schleicher, & Zilles, 1982) was a major step forward, since it provided a very fast and automated approach to the quantification of the major feature of cytoarchitecture, i.e. the regional-, laminar-, and cell column-dependent variation of cell packing density throughout the cortical ribbon (Amunts, Istomin, Schleicher, & Zilles, 1995…).” So, I have to repeat my complain: the technology published in 1995 by the group listed above, where my name was included, has nothing to do with GLI.

In the more recent paper, [4], with participation of the same authors (K. Amunts and K. Zilles) we read: “we were primarily interested in the percentage depth of layer IV, on which shrinkage would have little effect (Amunts et al, 1995)”. You probably guessed already that in the referenced paper the question of “shrinkage” was addressed in entirely different aspect, mainly – to justify profile’s depth normalization for the comparison of adult and children data ([5], page 567).  There were no discussion of shrinkage effect on position of peaks, especially – peaks of optical density profiles, since our profile registration was based on area fraction, neuronal sizes and numerical densities, and not staining intensity values. The irony of this particular reference is in the fact that referenced paper meticulously described an original algorithm that made registered profile independent from staining intensity as much as possible. Thus, it is quite wrong to use these results as an excuse to disregard the effect of tissue shrinkage on staining intensity profiles.

I could probably find more examples of the same referencing style, but first and foremost, an explanation is in order. Why should I feel so strongly about using the reference to this paper in the wrong context or for the wrong purpose? The answer is simple: the lion share of referenced publication is based on my dissertation entitled “Automatic Morphocorticography of human brain”, and it was developed quite independently, in Moscow, Russia (former USSR), without any help or advise from K. Zilles or A. Schleicher, who were working not only in quite another city – Düsseldorf, but in quite another country as well – Germany.

I started this work in late seventies – early eighties, and gradually transformed it into Ph.D. dissertation. It was defended in 1987, and all major results were published in Russia, between 1982 and 1987, in Moscow, USSR. Another part of the discussed paper is based on the dissertation of Katrin Fiedler (presently – prof. K. Amunts). She was working as a Ph.D. student in my group in the Laboratory of Clinical Neuroanatomy of the Moscow Institute of Psychiatry between 1982 and 1989. Working on her dissertation entitled “Quantitative analysis of the cytoarchitecture of Area 4 of human cerebral cortex in ontogenesis”, she used our technology for image processing, serial sections analysis and quantitative data collection. It was developed by me and my colleague, engineer Michael Shkliarov. At that time Katrin was working extremely hard, and demonstrated her best abilities, especially in tissue collection, processing and using statistical software for data analysis. She successfully defended her Ph.D. in Moscow in 1989. As you can guess, Prof. Zilles and Dr. Schleicher from the University of Düsseldorf could not possibly participate in the development of published technology, since at the time of its development and implementation we were separated, figuratively speaking, by two very strong walls: the “Berlin Wall” and the “Iron Wall”. In other words, wrong context notwithstanding, it is upsetting to be buried all these years behind “et al.” abbreviation, given the fact that the contribution of “co-authors”, listed in the title, requires very careful consideration, to say the least.

However, before doing exactly that, I have to mention another significant point: the text of the discussed paper where I supposed to be a co-author was produced without my participation whatsoever: before 2012 I did not even know about its existence.

2. The Content of the Papers 

Interested reader can use provided links to access the text of both papers. The first paper, entitled “Postnatal development of the human primary motor cortex: a quantitative cytoarchitectonic analysis” [5] is 15 pages long, including 2 and a quarter page of References. It has 16 pictures (some combined in pairs), numbered from 1 to 14, and 4 tables. This paper is the object of present discussion.

The second paper, where the same technology was used to analyse area 4 cytoarchitecture in Infantile Cerebral Palsy was entitled “Persistence of layer 4 in the primary motor cortex (area 4) of Children which Cerebral Palsy” (J. Brain Res., 38, 1997, 2, 247-260) [6]. I am not a co-author in this paper, so I am using it only to discuss the origin of brain tissue data, which is apparently the same in both papers.

3. The origin of the human brain tissue

The paper reports results of quantitative study of brain tissue taken on autopsies of subjects between 0 to 90 years of age, 21 adults (between 27 and 90 years of age) and 33 children (between 0 days to 13 years of age). Postmortem delay (time between death and autopsy) was specified as between 2 and 24 hours. The source of human tissue material is not specified. This omission implies that the tissue belongs to the institution of the corresponding author, namely – C. & O. Vogt Brain Research Institute, Heinrich Heine University Düsseldorf, Alternatively, the reader might assume that the tissue origin is the Mount Sinai Medical School in New York, what was the institution where I was working at the time of paper preparation. Still, neither of two options could not correspond to reality.

In 1989 K. Amunts (Fiedler), defended her dissertation [14], where she reported results of analysis of human brain cortex, also taken from individuals from 0 to 90 years of age, 20 adults and 36 children. Results were grouped by age in four group: group 1 – 0 to 2 years, group 2 – 3 to 6 years, group 3 – 7 to 14 years, and group 4 – 27 to 90 yeas of age. Postmortem delay was reported as between 2 and 6 hours for adult brains, and up to 24 hours for children. The cause of death for adults was cardiac death – “acute cardio-vascular insufficiency” or heart failure.

I know very well how the tissue, used in Katrin’s work was collected,  because between 1980 and 1986 I personally participated in collection of practically all adult brains, both “pathological” and of the “normal control”.  Approximately between 1980 and 1985 the Institute of Psychiatry participated in tissue collection program, administered by Moscow Institute of Cardiology (headed by prof. Chazov, who also was a Minister of Health).

The goal of the program was to collect the tissue of human heart to study mechanisms of cardiac arrest caused by acute vascular insufficiency. So, we had special permission of the Ministry of Health to perform postmortem study as soon as 2 hours after death, which was very important for morphological study of human brain as well. So, due to very short postmortem delay and the uniform cause of death (acute cardio-vascular insufficiency), our brain tissue collection was quite unique. These brains (35+ cases) were used as “normal control” in many research projects of the Institute of Psychiatry and Moscow Institute of Brain Research, including my dissertation, and the Ph.D. of Katrin Fiedler.

The source of children’s brains was different. The origin was Moscow Morosov’s Children’s Hospital; postmortem studies and the tissue was provided by Dr. V. Vinogradov, from Department of Pathology of this hospital (please, also see section 8 below). In this context it is important to consider a small paper [15], published by K. Fiedler in 1989 (no co-authors), where similar tissue collection was described. The paper described results of the study of 62 human brains between 0 and 90 years of age, 20 adults and 42 children. Postmortem interval was not specified. Age-related changes of the volume fraction of neurons in area 4 were reported for three age groups: group one – 0 to 1 year, VA = 0.1374, group two – 1 to 15  years,  VA = 0.0803, and group three – 27 to 90 years, 0.0740. Based on matching age interval and number of brains, we can conclude that the adult control group is the same, and children group is larger, but essentially the same as in Katrin’s Ph.D.

Matching previously published results with discussed papers made me think that cases published in 1995 paper were collected in Moscow. Therefore, before further analysis, I had to match all published cases with cases listed in K. Fielder’s PhD. dissertation [14], which copy is freely available by request from Central Research Medical Library (ЦНМБ) in Moscow.

As opposed to the second paper, the first did not present clinical information for described control cases. However, data tables from the second paper, as well as from K.Fiedler’s dissertation, list many details, like cause of death, neurological status, motor function etc. This makes case matching more reliable. So, for the purpose of case matching, case tables 1, 2 and 3 were used from the second paper as well. Results of this comparison are presented visually in pictures below (figures 1-4). To save space, some pages are presented partially. Full view of data tables from K. Fiedler’s Ph.D. are available using this link.

Fig. 1: Matching adult cases published in [5] to K.Fiedler’s PhD. cases [14], table 1.

Fig. 2 : Matching young control and age-matched control cases from paper [5] to paper [6].

Fig. 3: Matching Infantile Cerebral Palsy cases published in the second paper [6] to K.Fiedler’s PhD. table 2 [14], pages 53-57 (partial view).

Fig 4: Matching age-matched control from paper [6] to K.Fiedler’s PhD table 1 cases [14], pages 49-50 (partial view).

After my analysis I could make this conclusion: In the first paper 54 cases were published: 33 children and 20 adults (Table 1). The second paper used some of these cases as “young control” (ages 1 day to 1 year and 3 month), and “age matched control” (ages between 3 and 13 years and 1 month). Out of 17 cases of young control (ages 1 day to 1 year and 3 month), published in both papers, it was possible to match 14 cases to K. Fiedler’s PhD. by age and clinical details. Thee cases were mismatched by age data, which could be the result of a technical error, or it could be some “new” cases. Out of 16 cases of “age-matched control” all cases could be matched to K. Fiedler’s PhD.

Out of 20 “adult” control cases (ages 27 to 90 years) published in both papers all are matched to Katrin’s Ph.D. case tables. Four additional cases, used in Ph.D., were not used in these papers. Out of 14  Cerebral Palsy cases published in the second paper 11 are matched by age and clinical details. Two cases are “new” (were not included in Ph.D.) One case is a match by age and clinical details, but mismatched by siting ability (possible typo: “could sit up” in Russian text, “Could not sit up” in English text). Based on case code assigned to each case, which is just the date of postmortem study of each patient (see tables from Amunts’s Ph.D., fig.1-4 above), we can see that all normal control brains and Cerebral Palsy brains were collected between 1981 and 1988. The tabulated summary of every matched case is presented in a separate file: summary data tables link.

Overall, 61 out of 67 cases (17+16+20+14), or ~91%, published in two Karl Zilles’s papers in 1995 and 1997 can be matched by age and clinical details to cases collected, processed and quantitatively analysed in Moscow between 1981 and 1988, and included in my Ph.D. and/or in Ph.D. of Katrin Fiedler. This time period corresponds to dates when K. Amunts was working in the Institute of Psychiatry in Moscow, which was between 10 to 4 years before she started working in K. Zilles’s Institute in Düsseldorf in 1992 (or 1993). This can only indicate, that the lion share, if not all tissue material, used  in the discussed papers, originated not from Düsseldorf, Germany, but from sources in Moscow, USSR.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.