◇◇新语丝(www.xys.org)(xys.dxiong.com)(xys.3322.org)(xys.xlogit.com)◇◇

从国外引用情况分析魏于全Nature-Medicine论文的真实性

作者：lijun

　　先引用一下《人民日报》的报道：[今年３月３１日，记者当面向魏于全了
解论文的情况。魏于全几次提到，自己被司履生指出问题的两篇论文，在科学上
是站得住脚的。“目前，科学上的问 题，应该是说得很清楚了。因为我那个东
西是国际公认的。为什么说是国际公认呢？在我们这个基础上，又发表了一系列
的文章，国外做类似研究的有１７０篇，就是人家做了，引用我的。这种文章是
很多的。”魏于全表示，“国外没有哪一 个教授说反对我的，人家还在用。这
个领域现在非常火。” ]

　　因发表在Nature-Medicine一文是魏院士最重要的论文，我们就来看看它的
真实引用情况并就此作一简单分析。同时，本文将侧重点放在其特异性抗原上。

　　自2000 年发表至今，该文被国际期刊引用49次。其中，自引15次，综述文
章引了14次，研究性文章引用了20次。客观来说，这个引用记录还不错，可能与
发表文章的刊物有关。通常而言，研究人员更喜欢引用在高档次的杂志上发表的
文章。现在就看一下这20次研究性文章引用情况。除了有几篇与其文章有一定关
系外，大多数的引用仅仅只是在引言中提到一下而已，并非是继续或验证其工作。
事实上，除了魏院士再三提到支持其工作的两篇文章(下面会谈到)外，有的文章
不是重复不到魏的结果就是间接地否定他的结果，主要是在特异性抗原方面。现
就这些文章进行简单分析。【方舟子按：以下分析了5篇国外论文对魏论文的引
用情况（包括那两篇被魏认为是支持其结论的论文），结论是，目前没有人能够
重复出魏论文的结果，反而有一些论文得出了与魏论文截然不同的结论，因此魏
于全声称他的工作被国外论文引用、证实，是在误导同行和公众。】

1. This paper is the one mentioned the most, either positively as a 
support or negatively as a piece of evidence, by Professor Wei and 
other accusers. This paper was published by Scappaticci, FA et al. in 
2003 (Scappaticci, FA, et al. 2003, Polyclonal antibodies to 
xenogeneic endothelial cells induce apoptosis and block support of 
tumor growth in mice. 21:2667-2677). Previously, other people already 
pointed out that this paper cannot be used to support Wei’s work 
because of the differences in experimental design, results and 
mechanism. Now, let’s look at how Wei’s publication was actually 
cited in this paper. I quote here “Until now, there has been little 
evidence that endothelial cells or their products can be exploited for 
immunotherapeutic strategies in the tumor-bearing host [14]. Wei et al. 
have recently demonstrated that active vaccination of mice with human 
endothelial cells can break tolerance and induce a specific 
antiangiogenic immune response with broad antitumor activity [15]. 
While this strategy may result in a protective immune response for a 
variety of cancers, the use of endothelial cells to induce an immune 
response could potentially lead to toxic side effects. These include 
autoimmune responses against quiescent endothelium as well as 
inhibitory effects on angiogenesis during wound healing and 
revascularization. The use of passive immunotherapy, however, can be 
controlled more readily allowing the option of discontinuing therapy 
as required.” From this paragraph, Wei is the only group doing active 
vaccination with endothelial cells; even if Wei’s strategy works, 
there is high risk for therapeutic use; and the reason why passive 
immunotherapy is needed instead of active vaccination. I could’t see 
any supportive language in this paragraph. So, Wei’s claim of which 
his work was supported by this paper is false. In fact, there are many 
differences in the results compared to Wei’s paper as well: a. The 
antibody showed cross reactivity even to human non-endothelial cell 
types (DLD-1). I quote here “Although the avidity of the antibody was 
slightly greater, the results indicate cross-reactivity of the immune 
antibody to these divergent cell lines suggesting common antigenic 
epitopes on these cell lines (as would be expected if the targets 
recognized by these polyclonal antibodies were such molecules as 
histocompatibility molecules).” b. No specific antigen(s) could be 
detected and identified by Western blots, and the protein bands did 
not have any specificity. I quote here “Tumor cell lysates were then 
probed on Western blots with either pre-immune or immune IgG. Multiple 
bands were identified by immune IgG but not the pre-immune IgG. 
Futheremore, these bands were present in endothelial (SVR) and 
non-endothelial cell types (L1210, B16F10, and Phoenix A).” Therefore, 
rather than supporting Wei’s paper, their results were contradictory 
a lot to Wei’s results.

2. Second paper is the one published in 2004 (Corsini, E. et al. 2004, 
Immunotherapy with bovine aortic endothelial cells in subcutaneous and 
intracerebal glioma models in rats: effects on survival time, tumor 
growth and tumor necovascularization. Cancer Immunol. Immunother. 
53:955-962.). This paper has been cited as supportive to Wei’s work 
at many occasions. First of all, it would be suspicious if one has to 
use a paper published at a very low impact journal to support his/her 
work published at a high profile journal, and unfortunately this one 
is the case. Secondly, let’s look at how they cited Wei’s paper. I 
quote “A study by Wei [14] suggested that in a murine model, 
vaccination with actively proliferating xenogenic endothelial cells 
was effective in reducing tumor progression in mice bearing various 
types of tumor lines (i.e., Meth A fibrosarcoma, H 22 heptoma, MA 
782/5s, FM3A mammary carcinoma, and Lewis lung carcinoma). However, no 
glioma lines were used in the work by Wei et al [14], leaving the 
impact of this kind of treatment on glioma development unexplored.” 
This citation clearly indicated that their work was only a 
supplemental to Wei’s observation, which was absolutely not a 
validation of Wei’s study. Thirdly, although it’s not scientifically 
reasonable to compare rats with mouse models, it clearly indicated 
that there were significant difference between this paper and Wei’s 
paper. The efficacy of their endothelial cells was not as good as Wei’
s results as shown in Fig. 1 and Fig. 2 in which vaccinated tumor 
growth was inhibited only after day 17 and the difference between 
treated group and controls was not significant (Fig. 2). Finally, they 
were unable to identify any specific antigens causing the antitumor 
effects without any proteins like VEGFR II, alpha v integrin in Wei’s 
paper. I quote here ”By Western blot analysis, endothelial cell 
extracts showed two bands with molecular sizes of 11 and 19 kDa when 
probed with serum from rats with tumor immunized with bovine 
endothelium (Fig. 8),”. In fact, not only their results did not 
support Wei’s publication, but they also displayed different activity 
and molecular mechanism. Then, how can someone claim such results to 
support Wei’s paper? The only explanation is that the intention was 
to fool people without basic knowledge in biosciences.

3. The third paper cited Wei’s publication was published in 2004. 
(Okaji, Y. et al. 2004, Vaccination with autologous endothelium 
inhibits angiogenesis and metastasis of colon cancer through 
autoimmunity. Cancer Sci. 95:85-90). Based on Wei’s hypothesis, 
quoting from his paper “The breaking of immune tolerance to 
autologous angiogenic endothelial cells should be a useful approach 
for cancer therapy. However, immunity to angiogenic vessels is 
presumably difficult to clicit by autologous or syngeneic endothelial 
cells or their proteins as vaccine because of the immune tolerance 
acquired during the development of the immune system.”, it is 
unlikely to use own endothelial cells to produce antitumor activity. 
Unfortunately, this paper was directly contradictory to Wei’s 
hypothesis, and showed superior antitumor activity by vaccination of 
syngeneic endothelial cells, which resulted in greater antitumor 
activity compared to HUVES (Fig. 2). In addition, the immunoglobulins 
had no specificity. I quote here “Although the increase in serum 
concentration of immunoglobulins was small, in the ELISA assay, the 
antibodies showed a strong affinity for endothelial membrane antigens. 
In addition, the antibodies present in the sera of HSE-vaccinated mice 
strongly bound not only proteins of HSE membranes, but also those of 
HUVEC membranes (Fig. 4A), and vice versa (Fig. 4B). Thus, the 
antibodies induced by endothelial vaccines were cross-reactive with 
both murine and human antigens. Furthermore, the sera of mice of both 
groups did not react with membrane proteins of tumor cells (Fig. 4C), 
in accordance with the flow-cytometry analysis.” Now, let’s take a 
look what they cited from Wei’s paper. I quote here “Wei et al have 
reported that immunization with xenogeneic, but not autologous, 
endothelial cells, was effective in inducing an antiangiogeneic 
immunity and protecting mice from tumor growth... The present study, 
however, autologous endothelial vaccines were more active than the 
xenogeneic ones in inducing specific humoral and cellular immunities 
against tumor endothelium, and they consequently provided greater 
tumor inhibition.” Therefore, the difference is very obvious, which 
directly challenged Wei’s results, at least on the basis of his 
hypothesis of breaking tolerance of own immune system by xenogeneic 
antigens. Moreover, this paper was unable to identify specific 
antigens for the antitumor activity either, not mentioning about VEGF 
II and alpha v integrin.

4. It has been extensive interest in discovering novel 
anti-angiogeneic targets for therapies by both proteins and small 
molecules. The following paper was an attempt to identify new 
therapeutic target by establishing phage display technique in order to 
identify specific markers in endothelial cells. This paper was 
published by Mutuberria, R. et al. in 2004 (Mutuberria, R. et al. 2004, 
Isolation of human antibodies to tumor associated endothelial cell 
markers by in vitro human endothelial cell selection with phage 
display libraries. Journal of Immunological Methods, 287:31-47). First 
of all, let’s see what they say on Wei’s work. I quote here 
“Immunotherapy of tumors using fixed proliferating HUVEC as a vaccine 
has resulted in the inhibition of angiogenesis and regression of solid 
tumors (Wei et al., 2000)”. That’s the only place mentioned Wei’s 
work in the entire article. Since the same cells (HUVEC) were used and 
the molecular basis of antigens was explored in this paper, it’s 
quite reasonable to compare their results with Wei’s. Also, phage 
display is a powerful tool to probe specific peptide regions in the 
antigens. Any significant antigens would be picked up, such as the 
VEGFR-2 and alpha v integrin shown in Wei’s Fig. 7. However, in the 
results section, I quote here “We have used commercially available 
antibodies to compare these antigens with the molecular weights of 
possible target antigens such as CD31 (PECAM), Tie-1, Tie-2, VEGFR-1 
and VEGFR-2. None of the those commercial antibodies detected the 
exact same molecular weights as the selected phage antibodies (Fig. 2b 
and data not shown).” In fact, what they observed were the bands with 
molecular weights of 175 kDa and 110-125 kDa. Thus, the conclusion is 
that there must be somebody wrong, either the group in Netherlands or 
Professor Wei according to this paper.

5. Finally, let’s examine another paper published by Mittleman, A. et 
al. in 2002 on analysis of immune responses based on peptide 
similarity (Mittleman, A. et al. 2002, Monoclonal and polyclonal 
humoral immune response to EC Her-2/neu peptides with low similarity 
to the host’s proteome. International Journal of Cancer, 98:741-747). 
The intention of this paper was to rationalize the relationship 
between peptide sequences and host’s proteome in order to better 
predict the epitopic peptide sequence and corresponding immunogeneicity. 
In this paper, they cited Wei’s publication in the discussion as “In 
an effort to improve the ability of predicting disease relevant 
epitopic peptides, the present work has taken into consideration the 
self-character of TAAs and the matter of fact that immune tolerance 
appears as the main obstacle to an effective immune recognition and 
destruction of human cancers. 30-32”. Their results showed that there 
is a relationship between disease associated epitopes and the host 
proteome, but human brest/prostate cancer sera preferentially react 
with synthetic HER-2/neu peptides having low/medium level of 
similarity to human patients. In addition, the highest level of 
peptide immunogeneicity was shown in the peptides with no similarity 
at all to human proteome. Their conclusion was “We suggest that a low 
level of similarity to the host’s proteome might play a significant 
role in determining peptide immunogeneicity.”. In contrast, what Wei 
described in his paper and his theoretical basis of choosing peptides 
for synthesis and test their activity was “Sequence comparison 
analysis using the SwissPort database in NCBI indicated that the 
primary sequences of VEGFR II and alpha v integrin of mice and humans 
were homologs that were 82% and 89% identical, respectively, at 
amino-acid level. Next, we selected pairs of peptides for synthesis 
from the regions that shared the most-identical amino-acid sequences 
between humans and mice”. In his case, human peptides are immunogens 
whears mouse sequences represent the host proteome. Therefore, there 
is an obvious discrepancy between Wei’s theoretical basis and this 
paper’s experimental data.

In conclusion, to date, there has been no report on the detection and 
identification of VEGFR-II and alpha v integrin as antigens from 
endothelial cells when they were vaccinated in animals in literature. 
It’s fair to say that Professor Wei has mislead colleagues and the 
public on the actual citations of his paper published at 
Nature-Medicine in 2000. Nobody has been able to repeat his work in 
any scholastic publications so far, and the truth is that there are 
some publications that observed very different results and made 
opposite conclusions. If it were possible to repeat or verify his work, 
I would suggest to use his peptides listed in the paper to observe 
their antitumor activity, which is the simplest and most convenient 
way to draw a conclusion to determine whether he faked or manipulated 
the data in this very important publication.

(XYS20060416)

◇◇新语丝(www.xys.org)(xys.dxiong.com)(xys.3322.org)(xys.xlogit.com)◇◇