debsbb

I am going to make myself VERY unpopular here.
Firstly, there is no such antigen as Kell (the closest is K5 or Peltz), and there is no such antibody as anti-Kell (the closest is anti-K5 or Anti-Peltz).  The blood group system is Kell Blood Group System, but the first antigen within that system is K, and the antibody directed against that antigen is anti-K.  The ONLY individuals who can be described as being Kell Negative are those incredibly rare individuals who have the Ko phenotype.
Secondly, ONLY genes can be either homozygous or heterozygous (or, in some cases, hemizygous).  Antigens CANNOT be described as either homozygous or heterozygous (or, in some cases, hemizygous).  As red cells do not contain a nucleus, but merely express antigens, as the result of the individual inheriting certain genes, red cells CANNOT be either homozygous or heterozygous (or, in certain circumstances, hemizygous).  The correct way of describing such red cells is to describe them as having either homozygous expression, or heterozygous expression (or, in some cases, hemizygous expression).
Having got that off my chest (BUT, IT IS VERY IMPORTANT), I will now address the underlying question of this thread.
Without doubt, the reaction between an anti-K and the K antigen CAN show "dosage"; I don't think anyone would dispute that these days.  However, the more important question is, over the years and years and years that we have been using antibody screening cells that are K+k+, and have, therefore, NOT detected an anti-K that only reacts with red cells that have (supposed) homozygous expression, and do not forget, some of those cells may actually have come about as the result of a genotype of KE02/KEN, which means that the red cells actually have hemizygous (or "single dose"expression), how many patients have actually suffered a haemolytic transfusion reaction that has been clinically significant, causing morbidity or mortality?  I would contend that, having looked through as much of the available literature as I have been able, the answer is zero.
Obviously, if there is an anti-K present in the first place, albeit, it is undetectable by routine serological techniques, then transfusing the individual with either K+k+ or K+k- red cells will boost the anti-K, but it has not, as far as I know, resulted in mortality or morbidity, but will, almost certainly, result in an increase in titre and avidity of the antibody (I recognise, of course, that this could have disastrous results in a pregnancy, but only if the pregnancy is not properly monitored).
On top of all the above, it is well-known that there are mutations of the KEL1 gene, resulting in "unusual" amino acid substitutions, resulting in weakened expression of the K antigen, are not unknown.
It is well known that almost all blood group systems have their "quirks", which means that nobody can rely 100% on their antibody screening cells, BUT WE ALL DO!
I would politely suggest that an anti-K that is only detectable with red cells that are (genuinely) K+k- are not that important.