The online version of this article has a Supplementary Appendix. BackgroundClones of glycosylphosphatidylinositol-anchor protein-deficient cells are characteristic in paroxysmal nocturnal hemoglobinuria and are present in about 40-50% of patients with severe aplastic anemia. Flow cytometry has allowed for sensitive and precise measurement of glycosylphosphatidylinositol-anchor protein-deficient red blood cells and neutrophils in severe aplastic anemia. Design and MethodsWe conducted a retrospective analysis of paroxysmal nocturnal hemoglobinuria clones measured by flow cytometry in 207 consecutive severe aplastic anemia patients who received immunosuppressive therapy with a horse anti-thymocyte globulin plus cyclosporine regimen from 2000 to 2008. ResultsThe presence of a glycosylphosphatidylinositol-anchor protein-deficient clone was detected in 83 (40%) patients pre-treatment, and the median clone size was 9.7% (interquartile range 3.5-29). In patients without a detectable clone pre-treatment, the appearance of a clone after immunosuppressive therapy was infrequent, and in most with a clone pre-treatment, clone size often decreased after immunosuppressive therapy. However, in 30 patients, an increase in clone size was observed after immunosuppressive therapy. The majority of patients with a paroxysmal nocturnal hemoglobinuria clone detected after immunosuppressive therapy did not have an elevated lactate dehydrogenase, nor did they experience hemolysis or thrombosis, and they did not require specific interventions with anticoagulation and/or eculizumab. Of the 7 patients who did require therapy for clinical paroxysmal nocturnal hemoglobinuria symptoms and signs, all had an elevated lactate dehydrogenase and a clone size greater than 50%. In all, 18 (8.6%) patients had a clone greater than 50% at any given time of sampling. ConclusionsThe presence of a paroxysmal nocturnal hemoglobinuria clone in severe aplastic anemia is associated with low morbidity and mortality, and specific measures to address clinical paroxysmal nocturnal hemoglobinuria are seldom required.Key words: paroxysmal nocturnal hemoglobinuria, severe aplastic anemia. Haematologica 2010;95:1075-1080. doi:10.3324/haematol.2009 This is an open-access paper. Citation: Scheinberg P, Marte M, Nunez O, and Young NS. Paroxysmal nocturnal hemoglobinuria clones in severe aplastic anemia patients treated with horse anti-thymocyte globulin plus cyclosporine.Paroxysmal nocturnal hemoglobinuria clones in severe aplastic anemia patients treated with horse anti-thymocyte globulin plus cyclosporine

T imetabling the courses o †ered at the Computer Science Department of the University ofMunich requires the processing of hard and soft constraints. Hard constraints are conditions that must be satisÐed soft constraints, however, may be violated, but should be satisÐed as much as possible. T his paper shows how to model the timetabling problem as a partial constraint satisfaction problem and gives a concise Ðnite domain solver implemented with constraint handling rules that, by performing soft constraint propagation, allows for making soft constraints an active part of the problem ± solving process. Furthermore, efficiency is improved by reusing parts of the timetable of the previous year. T his prototype needs only a few minutes to create a timetable while manual timetabling usually takes a few days. T his was presented at the SystemsÏ98 Computer Fair in Munich and several universities have inquired about it.

Soft constraints are a generalization of classical constraints, which allow for the description of preferences rather than strict requirements. In soft constraints, constraints and partial assignments are given preference or importance levels, and constraints are combined according to combinators which express the desired optimization criteria. On the other hand, constraint handling rules (CHR) constitute a high-level natural formalism to specify constraint solvers and propagation algorithms. We present a framework to design and specify soft constraint solvers by using CHR. In this way, we extend the range of applicability of CHR to soft constraints rather than just classical ones, and we provide a straightforward implementation for soft constraint solvers.

Soft comtr~dnts are a generalizatkm of classical constraints, where constraints and/or partial n-'miLmments are associated to preferenco or importance levels, and constraints are combined according to combinators which expLe~ the demLred optimization criteria. Constraint I~anrllin~ RuleS (CHI~) constitute a high-level natural formalism to specify constraint sol,ram and propagation algorithrn~. In this paper we present a framework to design and specify soft coustraint solvers by using CHRs. In this way, we extend the range of applicability of CH]~ to soft constraints rather than just cl~ical ones, and we provide a straightforward implementation for soft constraint solver&

This paper takes three important steps towards constraint-based school timetabling: (i) It proposes a constraint model that covers many important requirements of school timetables by means of global constraints. (ii) It proposes a corresponding problem solver that learns from its earlier faults and restarts to escape non-promising parts of the search space. (iii) By reporting a large-scale computational study, it delivers a proof of concept.