Search

The aim of this thesis was to present the concepts of stationarity and reversibility in the modeling of the evolution of DNA nucleotide sequences, and to check whether they are valid for evolution of real genomes. To this end we have introduced the Stationarity Indices, STIs, which compare the current nucleotide distribution to the stationary one, and the Irreversibility Indices, IRIs, which are based on the Kolmogorov cycle conditions for the time reversibility of a Markov process. The indices can be easily computed once we have at disposal, using for example a maximum likelihood estimation, the rates of the process. We derived explicit expressions of the indices for the general 12 parameters model of nucleotide evolution with independent sites. It is interesting to note that assuming time-reversibility, which amounts to setting the IRI indices to zero, defines a 9-dimensional sub-manifold of the 12-dimensional space of all possible models. This manifold is the one spanned by the GTR model and its nested sub-models. We analyzed the analytical formulation of the indices for the reverse complement symmetric models. This particular parameterization arises in a natural way when describing evolution of neutrally evolving sequences. In this case it turns out that both STI and IRI have a simpler form. In particular one needs only one index, $\text{IRI}_{1}$, in order to test time reversibility. So imposing the constraint of time reversibility restricts the space of models to a 5-dimensional manifold in the 6-dimensional space of all the possible reverse complement symmetric models. We have successively extended the scope of our study to an evolutionary model which takes into account the $\CpG$ decay process, the predominant substitution process in vertebrates. This approach based on a set of indices is complementary to the one using a likelihood ratio test, and it has the advantage that it simultaneously assesses stationarity and time-reversibilty for all branches of a given phylogeny once the rate matrices have been estimated. On the contrary, a likelihood ratio test requires a comparison of different hypotheses on different branches and a new estimation of the parameters for each of them. When testing for all combinations the number of likelihood ratio tests required grows exponentially with the number of branches in the phylogeny. As an application of the theory we have measured the STI and IRI in two different species lineages, D. simulans and H. sapiens. Using a sliding window analysis and the maximum likelihood estimation method we have derived the distributions of $\text{STI}$ and $\text{IRI}_{1}$ for Drosophila, and of $\text{STI}$, $\text{IRI}_{1}$ and $\text{IRI}_{\text{\CpG}}$ for human. In both cases we find statistically significant deviations from equilibrium and time reversibility. In D. simulans, the values of $\text{STI}$ and $\text{IRI}_{1}$ are close to zero, suggesting that it is legitimate to use a time reversible Markov model in bioinformatics algorithms, for instance in those used for phylogenetic reconstruction. However, in the human lineage, we find substantial deviations from equilibrium and time-reversibility due to the CpG methylation deamination process, in particular $\text{IRI}_{\text{\CpG}}\approx 1$. In this case, the lack of equilibrium and time-reversibility is an important feature of the probabilistic model and consequently should not be disregarded., Ziel dieser Arbeit war es, die Bedingungen für Stationarität und Zeitreversibilität in Bezug auf die Modellierung der Evolution von DNS Sequenzen vorzustellen und zu überprüfen, ob diese Gegebenheiten bei der Evolution von genomischen DNS Sequenzen zutreffen. Zu diesem Zweck wurden Statinaritätsindices (STIs) die die derzeitige Nukleotidverteilung mit stationärer Nukleotidverteilung vergleichen, und Irrevesibilitätsindizes (IRIs), die auf Kolmogorovs Bedingungen für Zyklen zurückgehen, eingeführt. Diese Indizes können einfach errechnet werden, sobald die Raten des evolutionären Prozesses bekannt sind, z.B. durch eine Schätzung mittels Maximum Likelihood Verfahren. Es wurden explizite Ausdrücke für diese Indizes für das generelle 12 Parameter Modell der Evolution von DNS Sequenzen ohne Nachbarabhängigkeiten hergeleitet. Es ist interessant zu beobachten, dass unter der Annahme von Zeitreversibilität die drei IRIs verschwinden müssen und diese Bedingungen eine 9-dimensionale Untermannigfaltigkeit in dem 12-dimensionalen Raum aller Modelle aufspannen. Diese Untermannigfaltigkeit ist die des GTR Modells und aller seiner Untermodelle. Des Weiteren wurden diese Indizes für Modelle mit einer zusätzlichen Symmetrie, der reversen Komplementarität, die bei der Beschreibung von neutraler Evolution der doppelsträngigen DNS gegeben ist, hergeleitet. Unter dieser Symmetrie nehmen die Indizes eine einfachere Form an. Insbesondere gibt es nur noch einen Irreversibilitätsindex. Im zeitreversiblen Fall wird dadurch eine 5-dimensionale Untermannigfaltigkeit in dem 6-dimensionalen Raum der reversen komplementen Modelle beschrieben. Darüber hinaus wurden diese Konzepte auch für die Evolution von DNS Sequenzen mit Nachbarabhängigkeiten verallgemeinert, wie sie zum Beispiel durch den $\CpG$ Methylierungs- und Deaminationsprozess, der vor allem in Wirbeltieren ein sehr verbreiteter Mutationsprozess ist, entstehen. Dieser Zugang, die Stationarität und Zeitreversibilität anhand von Indizes zu prüfen, ist insbesondere bei großen phylogenetischen Bäumen einem Likelihood Ratio Test vorzuziehen, da er eine unabhängige überprüfung dieser Annahmen auf jedem Ast der Phylogenie zulässt. Ein Likelihood Ratio Test müsste demgegenüber alle möglichen Kombinationen berücksichtigen und deshalb exponentiell viel häufiger ausgeführt werden. Im Rahmen einer Anwendung unserer theoretischen überlegungen, wurden die IRIs und STIs für die Nukleotidevolution in der menschlichen Linie (Homo Sapiens) und in der Fruchtfliege (Drosophila Simulans) berechnet. Die Indizes wurden in verschiedenen Regionen aus den Mutationsraten berechnet, welche mittels Maximum Likelihood Methode gemessenen worden waren. In beiden Spezies fanden wir statistisch signifikante Abweichungen der Stationarität und Zeitreversibilität. In der Fruchtfliege sind die Abweichungen klein und die Verwendung von bioinformatischen Methoden, die diese Annahmen machen, erscheint legitim. In der menschlichen Linie allerdings sind die Abweichungen substanziell größer, was zuallererst auf die $\CpG$ Methylierung und Deamination zurückzuführen ist. In diesem Falle ist das Nichtvorhandensein von Stationarität und Zeitreversibilität eine Tatsache, die bei der statistischen Beschreibung und Modellierung nicht vernachlässigt werden sollte.

p53 is a tumor suppressor gene, located in the short arm of chromosome 17, which encodes for a nuclear protein involved in the control of cellular growth. Mutations in p53 gene are the most common genetic alterations in a several human cancers, including Non Small Cell Lung Cancer (NSCLC). However, up to now, the role of p53 in the tumour's behaviour and its progression has not been completely clear. We performed immunohistochemical staining for mutated p53 using two monoclonal antibodies, PAb1801 and PAb240, in fresh tumour specimens from 103 consecutive patients who underwent surgery for resectable NSCLC. PAb1801 detects both the normal and mutant form of p53, while PAb240 is specific only for the mutant form and recognizes a denaturation-resistant epitope located between aminoacids 156-335. Both antibodies showed a mainly nuclear staining in neoplastic cells but not in surrounding uninvolved lung tissues. 68 out of 100 (68%) and 37 out of 103 (35.9%) of the cases were positive with PAb1801 and with PAb240, respectively. Tumours from patients with hilar-mediastinal lymph node involvement showed a higher p53 expression, detected by PAb1801, than those without nodal metastases (p = 0.04). Moreover, tumours expressing more than 60% of positive cells with both antibodies showed a significant increase of nodal involvement (p = 0.1; p = 0.03). Furthermore, p53 expression was significantly related to post-surgical stage (p Tumor Stage) (p = 0.04). In addition, we did not find any correlation between p53 expression and proliferating activity evaluated by PCNA, Ki-67 and DNA flow cytometric cell cycle. In conclusion, the evaluation of p53 oncogene expression may identify individuals whose resectable NSCLCs have a more aggressive tumour behaviour.

Twenty-eight patients with small cell lung cancer were treated with high dose 4'-epidoxorubicin (EDX). Fifteen patients underwent EDX monotherapy (cumulative dose: 800 mg/m2) while 13 were treated with EDX (cumulative dose: 660 mg/m2) associated with cyclophosphamide, etoposide and cisplatin. A 2D-echo was performed in basal condition, after the third and sixth dose and 2 months after the end of the therapy. In the patients without cardiovascular disease (n = 19) left ventricular end-diastolic volume (EDV), end systolic volume (ESV) and ejection fraction (EF) were unchanged. The patients with coronary artery disease (n = 5) showed a statistically significant decrease in EF with an increase of ESV. All patients with systemic hypertension (n = 4) showed a significant reduction of EF and a significant increase of ESV and EDV.

The sperm collected from mammary tumor virus (MTV)-carrying mice (C3H, BALB/cfC3H, BALB/cfRIII and RIII) was separately tested for mammary tumor-inducing activity in (BALB/c × C3Hf)F1, (BALB/c × RIIIf)F1, and BALB/c female recipients by i.p. injection of 0.1 ml of the sperm at 1–2 weeks or at 3 months of age. A total of 551 recipients was observed, including control mice. The results may be summarized as follows: 1) mammary tumor incidence in experiments with or without histocompatibility between sperm donor and recipient is the same; 2) bioactivity is related to the type of MTV (C3H, RIII) and to the type of recipient, not to the sperm donor; 3) the activity of RIII MTV released in the sperm appears to be less influenced by the age of recipients than is that of C3H MTV; 4) BALB/c recipients are more susceptible to C3H than to RIII sperm-released MTV; 5) (BALB/c × RIIIf)F1 hybrids are resistant to sperm-released MTV, especially to C3H MTV infection, and show a 34% incidence of late spontaneous lymphomas inherited by the RIIIf male parent; 6) (BALB/c × C3Hf)F1 hybrids are susceptible to both C3H and RIII sperm-released MTV and show a 30% incidence of late spontaneous mammary tumors due to genetic transmission of MTV by the C3Hf male parent.

Monoclonal antibodies B 6.2, B 72.3 and B 1.1, reactive against tumor-associated antigens of human breast cancer, have been tested on histologic sections of paraffin blocks from 49 cases of breast cancer and from 13 cases of mammary dysplasia. The results have shown that: (a) the three monoclonal antibodies recognize different antigenic determinants of the cancer cells; (b) monoclonal antibodies B 6.2 and B 1.1 recognize antigens which are frequently expressed in cancer cells; (c) the antigen recognized by monoclonal antibody B 72.3, at the concentrations used, is less frequently expressed in tumors, while it seems to be selective for the apical cell border and luminal secretions of apocrine metaplastic lobules or cysts occurring in mammary dysplasia; (d) intracytoplasmic expression of the antigens is observed only in cancer cells, with rare exceptions; (e) tumor cell populations are heterogeneous in respect to the expression of the antigens detected by the monoclonal antibodies tested; (f) the antigens detected are unrelated to the tumor histotype; (g) the frequency of axillary lymph nodal metastases is not influenced by the presence or intensity of antigenic expression in the primary tumors; but (h) in metastatic tumor cell populations the antigenic expression is higher than in primary tumors. This fact suggests some relationship between tumor metastatization and the expression of antigens recognized by the monoclonal antibodies tested.

In order to more clearly define the factors influencing prognosis, 113 cases of T1-T2N0 breast cancers, all operated on by means of radical mastectomy, were reviewed. The following pathologic parameters were analyzed in relation to 5 years survival: histological type, nuclear grading, vascular embolization, size and border of the tumour, and sinus histiocytosis, diffuse cortical hyperplasia and germinal center hyperplasia of axillary nodes. We observed a close correlation between the presence of sinus histiocytosis and prognosis, as many authors have reported. Diffuse cortical hyperplasia stands out from our data as the most unfavourable prognostic factor.