Reviewer comments until Section 4.4

Chapters/CASCADE/cohort.tex

@@ -3,11 +3,11 @@
 \section{Cohort level analysis}
 \label{cascade-sec:cohortLevel}
 
-Even though there were only five lung cancer patients who have participated in the CASCADE program to date, each of the patients had a high number of samples analysed, revealing the complexity of intra-patient heterogeneity in NSCLC (\autoref{cascade-sec:patientLevel}). Despite this heterogeneity, there were several parallels between the patients showing similarities in disease trajectories. The process of small cell transformation (SCT) is still significantly under explored and understood, due to the rarity of the transformation as well as the lower overall survival in comparison to other resistance mechanisms. In the following section, the patients who developed small cell carcinoma transformation were compared and contrasted with the adenocarcinoma cases to further explore this mechanism of resistance.
+Even though \remove{there were} only five lung cancer patients \remove{who} have participated in the CASCADE program to date, each of the patients had a high number of samples analysed, revealing the complexity of intra-patient heterogeneity in NSCLC (\autoref{cascade-sec:patientLevel}). Despite this heterogeneity, there were several parallels between the patients showing similarities in disease trajectories. The process of small cell transformation (SCT) is still significantly under explored and understood due to the rarity of the transformation \change{as well as}{and} the lower overall survival \change{in comparison}{compared} to other resistance mechanisms. In the following section, the patients who developed small cell carcinoma transformation were compared and contrasted with the adenocarcinoma cases to \remove{further} explore this mechanism of resistance\add{ further}.
 
-The generally accepted hallmarks of SCT, apart from the histological changes such as high \textit{MKI67} expression and down regulation of major histocompatibility complex I and II, are a much higher mortality, a high prevalence of \textit{FHIT} and \textit{MAD1L1} deletions or loss, and \textit{TP53} and \textit{RB1} mutations \cite{Meerbeeck2011,Raso2021}. However, while in patient CA-L all samples showed a TP53~``stop gained`` mutation, patient CA-I's transformation did not occur in the setting of  \textit{TP53} loss. Additionally, neither of the patients presented with a \textit{RB1}, \textit{FHIT}, or \textit{MAD1L1} loss (\Autoref{fig:ca51heatmap,fig:ca86heatmap}). 
+The generally accepted hallmarks of SCT, apart from the histological changes such as high \textit{MKI67} expression and down-regulation of major histocompatibility complex I and II, are \remove{a} much higher mortality, a high prevalence of \textit{FHIT} and \textit{MAD1L1} deletions or loss, and \textit{TP53} and \textit{RB1} mutations \cite{Meerbeeck2011,Raso2021}. However, while in patient CA-L, all samples showed a TP53~``stop gained`` mutation, patient CA-I's transformation did not occur in the setting of  \textit{TP53} loss. \change{Additionally, n}{N}either of the patients presented with an \textit{RB1}, \textit{FHIT}, or \textit{MAD1L1} loss (\Autoref{fig:ca51heatmap,fig:ca86heatmap}). 
 
-In agreement with recent literature showing whole genome doubling (WGD) for SCLC \cite{Zhou2021}, we observed chromosomal arm amplification in patient CA-I and full WGD for patient CA-L. However, all NSCLCs patient also showed at least one sample with complete WGD, casting doubt on WGD being a distinguishing feature of SCT (\Autoref{tab:ca99cnv,tab:ca51cnv,tab:ca80cnv,tab:ca82cnv,tab:ca86cnv}). Additionally, the overall loss of heterozygosity could be seen in both NSCLC and SCLC and this seems to be a general feature of late stage lung cancers rather than NSCLC (\autoref{fig:cascadeLOH}) suggesting that copy number alterations are not the main drivers of SCT.
+In agreement with recent literature showing whole genome doubling (WGD) for SCLC \cite{Zhou2021}, we observed chromosomal arm amplification in patient CA-I and full WGD for patient CA-L. However, all NSCLCs patient also showed at least one sample with complete WGD, casting doubt on WGD being a distinguishing feature of SCT (\Autoref{tab:ca99cnv,tab:ca51cnv,tab:ca80cnv,tab:ca82cnv,tab:ca86cnv}). Additionally, the overall loss of heterozygosity could be seen in both NSCLC and SCLC\change{, and this seems}{. This tendency towards anuploidy and heterozygosity seemed} to be a general feature of late-stage cancers rather than just NSCLC (\autoref{fig:cascadeLOH}, \cite{Girish2023}\note{added new publication}), suggesting that copy number alterations are not the main drivers of SCT.
 
 \begin{figure}[ht]
 \centering
@@ -16,9 +16,9 @@ \section{Cohort level analysis}
 \end{figure}
 
 
-The most prominent difference of NSCLC and SCLC in our patients was the reconstructed phylogeny. While the NSCLC showed substructure and meaningful evolutionary splits, the SCLC patients phylogenies showed a distinct ``star shaped`` pattern. Each sample branch was very close to the others and with substantial amounts of private variants in each sample (\Autoref{fig:CA51mitoPhylo,fig:CA86mitoPhylo} vs. \Autoref{fig:CA99mitoPhylo,fig:CA80mitoPhylo,fig:CA82mitoPhylo}). Even though the shared variants in CA-L seemed to provide the ability to transform, they do not necessitate the transformation, as both samples CA-L 17A and 26 remained NSCLC with virtually no known genetic determinant of status. This in term suggested either a currently undetected genetic determinant or potential epigenetic regulation to explain the SCT (\autoref{fig:ca86heatmap}).
+The most prominent difference between NSCLC and SCLC in our patients was the reconstructed phylogeny. While the NSCLC showed substructure and meaningful evolutionary splits, the SCLC patients' phylogenies showed a distinct ``star-shaped`` pattern. Each sample branch was very close to the others\change{ and}{,} with substantial amounts of private variants in each sample (\Autoref{fig:CA51mitoPhylo,fig:CA86mitoPhylo} vs. \Autoref{fig:CA99mitoPhylo,fig:CA80mitoPhylo,fig:CA82mitoPhylo}). Even though the shared variants in CA-L seemed to provide the ability to transform, they do not necessitate the transformation, as both samples CA-L 17A and 26 remained NSCLC with virtually no known genetic determinant of status. This\add{ lack of genetic change}, in turn, suggested either a currently undetected genetic determinant or a \change{potentially}{conceivable} epigenetic regulation to explain the SCT (\autoref{fig:ca86heatmap}).
 
-In contrast to CA-L, who presented with both NSCLC and SCLC, CA-I's samples were completely transformed. The biopsied adenocarcinoma, which already had an MHC-I disrupting mutation was completely out-competed by a secondary clone, which did not present with any additional genetic driver alterations. Similar to patient CA-L this suggested, that the genetic prerequisites for SCT were already present in the clonal population, but not sufficient to drive transformation (\Autoref{fig:ca51.clonalTree,fig:ca51.ccfCluster}). 
+In contrast to CA-L, who presented with both NSCLC and SCLC, CA-I's samples were completely transformed. The biopsied adenocarcinoma, which already had an MHC-I disrupting mutation, was completely out-competed by a secondary clone, which did not present with any additional genetic driver alterations. Similar to patient CA-L this suggested that the genetic prerequisites for SCT were already present in the clonal population but not sufficient to drive transformation (\Autoref{fig:ca51.clonalTree,fig:ca51.ccfCluster}). 
 
-Gene fusions or regulatory genetic variants leading to aberrant transcription could have been the cause for this phenomenon observed in both SCT cases. These could be detected in RNA sequencing of the biobanked fresh autopsy samples to exclude genetic causes which were not picked up with the performed WES, or detect transcription alterations. However, this analysis is outside the scope of this work.
+Gene fusions or regulatory genetic variants leading to aberrant transcription could have been the cause for this phenomenon observed in both SCT cases. These could be detected in RNA sequencing of the biobanked fresh autopsy samples to exclude genetic causes which were not picked up with the performed WES or detect transcription alterations. However, this analysis is outside the scope of this work.
 

Chapters/CASCADE/mitochondrial.tex

@@ -104,4 +104,4 @@ \subsubsection{Patient CA-L}
 \subsection{Summary}
 With the analysis of the mitochondrial history of samples, we could shed some light on the timing of lesions and the development of resistance mechanisms, which is not heavily influenced by the treatment and its selection pressure. While the infinite sites hypothesis does not hold true for mitochondrial DNA, due to the limited sites and reduced repair mechanisms, the selection pressure of treatment and their resistance mechanisms parallel evolution bias in the analysis of multiple related tumour samples also violates multiple assumptions for phylogenetic reconstruction when using somatic variants. 
 
-\note{massive rewrite}Neither options come without pitfalls and caveats, but this method could offer an alternate view on the history and seeding time of lesions and their kinship using data that was previously discarded but was abundantly available. This approach is also available at virtually no additional cost, as mitochondrial variants can be readily detected from standard WGS and WES. As mitochondrial DNA analysis has recently been moved into the spotlight as a biomarker for stress and mortality \cite{Trumpff2021} and other diseases \cite{Cushen2022} as more than just ``the power house of the cell``, we were interested similar to circulating mitochondrial DNA, tissue mitochondrial DNA could allow another dimension. And while certain aspects of evolutionary trajectory and patterns are conserved between the nuclear DNA and the mitochondrial DNA, many are not. These differences might be caused by the reduced granularity of the mitochondrial analysis, where the median number of somatic variants was 345 (min: 128, max 5401} over the median of 23252 (min: 10947, max: 64580) in the nuclear DNA, however they could also be caused by a different biological process. The method is not an alternative to nuclear DNA phylogeny, but rather points out a need for further investigation. The reconstruction methods should be approximately equivalent with the current biological insight. Furthermore, mitochondrial DNA was successfully used to distinguish clones in single-cell sequencing \cite{Ludwig2019}. This disconnect warrant further investigation of mitochondrial evolution in the cancer setting, outside the scope of this work.
+\note{massive rewrite}Neither options come without pitfalls and caveats, but this method could offer an alternate view on the history and seeding time of lesions and their kinship using data that was previously discarded but was abundantly available. This approach is also available at virtually no additional cost, as mitochondrial variants can be readily detected from standard WGS and WES. As mitochondrial DNA analysis has recently been moved into the spotlight as a biomarker for stress and mortality \cite{Trumpff2021} and other diseases \cite{Cushen2022}, we were interested similar to see if tissue mitochondrial DNA analysis, like circulating mitochondrial DNA, could unlock another multi-omics dimension. And while certain aspects of evolutionary trajectory and patterns are conserved between the nuclear DNA and the mitochondrial DNA, many are not. These differences might be caused by the reduced granularity of the mitochondrial analysis, where the median number of somatic variants was 345 (min: 128, max 5401} over the median of 23252 (min: 10947, max: 64580) in the nuclear DNA, however they could also be caused by a different biological process. The method is not an alternative to nuclear DNA phylogeny, but rather points out a need for further investigation. The reconstruction methods should be approximately equivalent with the current biological insight. Furthermore, mitochondrial DNA was successfully used to distinguish clones in single-cell sequencing \cite{Ludwig2019}. This discrepancy of results from nuclear and mitochondrial DNA phylogenetic reconstruction, warrants further investigation of mitochondrial evolution in the cancer setting, outside the scope of this work.

Chapters/CASCADE/outlook.tex

@@ -1,8 +1,8 @@
-\section{Outlook}
+\section{Discussion}
 \label{cascade-sec:outlook}
 
 In this chapter we described both the high inter and intra patient heterogeneity of late stage lung cancer patients and showed that mitochondrial variant based phylogeny could help to resolve the sample relationships in the context of selection pressure through treatment. Additionally we uncovered a different disease trajectory for cases showing SCT where the cancers appeared genetically primed for SCT but genetic primed, but genetic alterations alone were on sufficient to drive transformation. This suggests that genetic analysis alone will not allow the prediction or early detection of SCT as a resistance mechanism. This uncoupling of genetic evolution and disease histology was also reported recently in a study focusing on multi-region analysis of treatment na\"ive SCLC cases~\cite{Zhou2021}.
 
-While there exist several multi-region lung cancer studies up to date, their focus was on early stage and treatment na\"ive disease \cite{JamalHanjani2017,Leong2018}. While these studies showed ubiquitous intra-tumour heterogeneity and copy number alterations in early stage disease, there is an unmet need for the assessment of late stage lung cancers.
+While there exist several multi-region lung cancer studies up to date, their focus was on early stage and treatment na\"ive disease \cite{JamalHanjani2017,Leong2018}. While these studies showed ubiquitous intra-tumour heterogeneity and copy number alterations in early stage disease, there is an unmet need for the assessment of late stage lung cancers\add{, especially with the ability of a full body work-up and not just limited to surgically resected specimen }\cite{Zhang2014}. \add{ As surgical specimen specimen provide very high quality samples, almost all large scale studies rely on these early-stage lung cancers}\cite{Network2012,CGARN2014,George2015}\add{, which is still considered ground truth for tumour heterogeneity in lung cancer. However this lack of insight into in advanced and late-stage cancer heterogeneity may affect the development of effective target therapies for these cancers.}
 
 With this work we took a first step towards understanding and measuring the heterogeneity of tumour samples and treatment resistance mechanisms in late stage NSCLC, but many unanswered questions remain. The epigenetic marks and their inheritance patterns in cancer are a massive unexplored field which increases the heterogeneity of cancer even more \cite{Easwaran2014}. Additionally, we could only hypothesise and reconstruct the longitudinal trajectory of the disease from autopsy samples. The next step to validate and further explore these findings would be to analyse temporally spaced samples from the same disease.