A single-cell functional precision medicine landscape of multiple myeloma

Plasma cells are antibody producing cellular factories that play an essential role in the immune system. Defects in the development of plasma cells can cause severe diseases, including autoimmunity and cancer. For example, malignant plasma cells (also called myeloma cells) can build up in the bone marrow, leading to multiple myeloma (MM). MM is the second most common hematological malignancy, primarily affecting the elderly. It initially presents with few symptoms, but as the disease progresses, MM can lead to hypercalcemia, renal insufficiency, anemia, bone destruction, immunological impairment, and death. In the past two decades, MM patients have benefited from various novel treatment options, improving their overall survival. Despite this progress, patients still face a median 5- year survival rate of around 50%, in part due to drug resistance, escape and subsequent relapse of the disease. At relapse, the genetic complexity of MM is increased. Combined with the ever-growing set of approved treatment options, treating physicians face a real challenge in choosing the ideal therapy for an individual patient. Therefore, we need better ways to tailor and individualize MM treatment strategies. Precision medicine aims to match patients with treatments targeting their specific disease. Successful examples of precision medicine use genome sequence information to tailor patient treatment. However, in MM, this has had limited success due to the genetic and molecular complexity of the disease. A complementary approach exposes cells from a patient to possible treatment options such as drugs and immunotherapies outside of the body (ex vivo) and measures the response to identify effective treatment options. Such ‘functional precision medicine’ readouts might capture and integrate several levels of complexity of the disease: the genome, epigenetic and transcriptional regulation, the proteotype, metabolic fluxes, and cellular interactions and environmental influences. Several methods can be used to measure such ex vivo drug responses. Imaging by automated microscopy provides a highly scalable, clinically applicable method to measure cellular function and morphologies in response to tested drugs. This thesis aims to improve our understanding of this complex disease and bring functional precision medicine in multiple myeloma, based on automated microscopy, closer to the clinic. First, we develop a single-cell ex vivo drug screening approach called Pharmacoscopy for the MM setting. Then, we use Pharmacoscopy to systematically measure the response to targeted drugs and immunotherapies on 138 separate MM samples. By integrating Pharmacoscopy with patient characteristics and large-scale molecular measurements (genetics and proteomics), we learn more about the disease and evaluate how this single-cell functional precision medicine platform can aid the treatment of MM patients. Chapter 1 introduces multiple myeloma, its developmental origins, and current treatment strategies. The chapter finishes on current large-scale molecular measurements (OMICs) results and their contribution to precision medicine in MM so far. Chapter 2 presents the results of an integrative analysis centered around Pharmacoscopy measurements in a cohort of 138 MM samples. First, we introduce the method and graphically represent molecular and clinical diversity in our MM patient cohort (Figure 1). Next, we identify reoccurring modes in the morphological and cellular heterogeneity of MM samples, which we call PhenoGroups. We show that these PhenoGroups represent different disease stages and genetics, with distinct morphological and inflammatory signatures (Figure 2). We morphologically define and validate the detection of myeloma cells using convolutional neural networks (Figure 2) and validate their associated molecular signature by integration and validation using proteotyping and genetic profiling (Figure 3). Furthermore, we compare the protein-level molecular signatures of myeloma cells to a recently published single-cell RNA sequencing dataset. Thus, we show that our novel myeloma cell detection approach, and the resulting molecular profiling, is clinically relevant and a solid basis for pharmacoscopy (Figures 2-3). In Figure 4, we present the ex vivo drug responses to 61 existing and novel therapeutic strategies, including combinations of small compounds and antibody-based immunotherapies. The resulting drug response map reveals considerable differences in how patients respond ex vivo to the same treatments. By integrating drug responses with proteotypes, we reveal and analyze the molecular network behind these variable drug responses. The approach recovers known regulators of drug response in myeloma and identifies new modes by which the response to a drug can be molecularly regulated. In Figure 5, we integrate patient genetics and other characteristics to propose novel guidelines for the individualized treatment of patients. We show that the PhenoGroups are a strong indicator of ex vivo drug sensitivity, exemplifying the importance of morphologies and the cellular bone marrow composition in determining drug responses. In Figure 6, we use longitudinal clinical follow-up of patients and show that Pharmacoscopy identifies patients with prolonged clinical benefit for immunotherapy treatments, giving confidence for successful translation to the clinic. We further show preliminary data that suggests that inflamed PhenoGroup signature might also be helpful in the identification of MM patients that might have reduced benefit from immunotherapy. Lastly, in Chapter 3, we address the limitations of the work, discusses open-analytical avenues and proposes ways to expand on this work - on Pharmacoscopy specifically, with a focus on both experimental and clinical future applications. In summary, the results of this work are two-fold: We show that Pharmacoscopy can be directly-applicable to guiding patient treatment decisions for multiple myeloma. And, our integrative functional, molecular, and clinical analysis of primary myeloma samples advances our understanding of this complex disease