R-IPI Calculator for DLBCL
R-IPI Calculator for DLBCL
The revised IPI changes no factor and no weight. It redistributes the same five points of the 1993 index into three strata instead of four, because R-CHOP collapsed the difference between the two middle groups the original index separated.
R-IPI (Sehn 2007)
5 factors → 3 strataAge 68; LDH 420 U/L against a laboratory upper limit of 250 U/L; ECOG 1; Ann Arbor stage II; one extranodal site
The same five factors, regrouped
One point each; range 0 to 5
R-IPI (Sehn 2007): very good 0 · good 1–2 · poor 3–5
1993 IPI (Shipp): low 0–1 · low-intermediate 2 · high-intermediate 3 · high 4–5
- nothing was reweighted
- the R-IPI uses the identical five factors with the identical single point each. The only change is where the lines between strata fall. That is unusual among revised scores and it is the thing to understand about this one: it is a regrouping, not a refitting
- why four groups became three
- in R-CHOP-treated patients the 1993 index’s low-intermediate and high-intermediate groups no longer had distinguishable outcomes, while the pooled 0-or-1 low-risk group contained a genuinely excellent subset that the original could not isolate. Splitting 0 from 1 and merging 3 with 4–5 fixed both at once
- the LDH factor is a comparison with the local interval
- ‘LDH above normal’ means above the REPORTING LABORATORY’s upper limit of normal. This page takes the measured value and the limit separately so the comparison is visible, and because the NCCN-IPI bands the same ratio rather than dichotomising it
- the extranodal item differs from the NCCN-IPI’s
- here it is a COUNT, positive above one site. On the NCCN-IPI it is a NAMED LIST of four sites, positive if any one of them is involved. The two can disagree in both directions on the same patient
- the era
- the factors are from the 1993 International Prognostic Index, derived before rituximab in patients treated with anthracycline-based chemotherapy, where the four groups’ five-year overall survival was 73%, 51%, 43% and 26%. The REGROUPING was derived in a population-based British Columbia cohort treated with R-CHOP, which is why its own figures (94%, 79% and 55% at four years) are so much better than the 1993 numbers for the same factor counts
Worked example
Age 68; LDH 420 U/L against a laboratory upper limit of 250 U/L; ECOG 1; Ann Arbor stage II; one extranodal site
LDH ratio = 420 / 250 = 1.68, which is above 1, so 1 point
Age 68 is over 60, so 1 point. ECOG 1, stage II and a single extranodal site score nothing
1 + 1 + 0 + 0 + 0 = 2 points
2 sits in the R-IPI's 1–2 good stratum: four-year progression-free survival 80% and overall survival 79% in the R-CHOP derivation cohort
The same 2 points is low-intermediate risk on the 1993 four-group scheme, whose five-year overall survival for that group in the pre-rituximab cohort was 51%. Identical factors, identical points, two different labels and a 28-point difference in the quoted survival — almost all of which is rituximab rather than any disagreement about the arithmetic
Add stage III–IV and the total becomes 3: POOR on the R-IPI, high-intermediate on the original
Now change only the laboratory. At an upper limit of 450 U/L the same measured 420 is a ratio of 0.93, the LDH point disappears and the total falls to 1 — the bottom of the good stratum. Doubling both numbers together, to 840 against 500, leaves the ratio at 1.68 and the total at 2
The same five points, two groupings, two eras
| Points | 1993 IPI group | 5-year OS, pre-rituximab | R-IPI group | 4-year OS, R-CHOP |
|---|---|---|---|---|
| 0 | Low | 73% | Very good | 94% |
| 1 | Low | 73% | Good | 79% |
| 2 | Low-intermediate | 51% | Good | 79% |
| 3 | High-intermediate | 43% | Poor | 55% |
| 4 – 5 | High | 26% | Poor | 55% |
Three revisions of the same index, and what each changed
| Index | What it changed | Strata | Needs the local LDH limit? |
|---|---|---|---|
| IPI (1993) | The original five factors | 4 (0–1, 2, 3, 4–5) | Only as a yes/no |
| R-IPI (2007) | Regrouped the same points for the R-CHOP era | 3 (0, 1–2, 3–5) | Only as a yes/no |
| NCCN-IPI (2014) | Banded age into four and the LDH ratio into three; replaced the extranodal count with a named list of four sites | 4 (0–1, 2–3, 4–5, 6–8) | Yes, as a banded ratio |
A regrouping, not a refitting, and what rituximab did to the middle
The revised International Prognostic Index is the most economical revision in this category: it changes no variable, no threshold and no weight. The five factors are the 1993 index’s — age over 60, LDH above the laboratory’s upper limit of normal, ECOG performance status 2 or worse, Ann Arbor stage III or IV, and more than one extranodal site — each worth one point, total 0 to 5. All the R-IPI does is draw the lines between strata in different places: 0 alone, then 1 to 2, then 3 to 5.
That was the right change because of what R-CHOP did to the middle of the distribution. In the pre-rituximab cohort the 1993 index’s four groups had five-year overall survival of 73%, 51%, 43% and 26% — a usable spread. Once rituximab was added to CHOP, the low-intermediate and high-intermediate groups stopped being distinguishable from each other, so separating them bought nothing; and at the same time the pooled low-risk group turned out to contain a subset with an outcome the original index had no stratum good enough to describe. Splitting 0 from 1 and merging 3 with 4 to 5 solved both problems with no new data collection at all. In the derivation cohort the three R-IPI strata had four-year overall survival of 94%, 79% and 55%.
The practical consequence is a naming hazard. A total of 2 points is low-intermediate risk under the original scheme and GOOD under the R-IPI; a total of 3 is high-intermediate under the original and POOR under the R-IPI. The arithmetic is identical, so a letter or a database field recording ‘IPI: high-intermediate’ or ‘poor risk’ without naming the index has lost the information that matters. The same trap repeats one level up: the NCCN-IPI, published in 2014, scores the same five variables on an eight-point scale with banded age and a banded LDH ratio, and its four groups carry the same four names as the 1993 index’s while meaning different score ranges.
Two items deserve care. The LDH factor is a comparison with the reporting laboratory’s own interval, and this page takes the measured value and that limit separately rather than asking for a yes/no — partly so the comparison is visible, partly because the NCCN-IPI bands the same ratio and the two pages should not disagree about what it is. The extranodal item is a COUNT here, positive above one site, where the NCCN-IPI uses a named list of four sites; the two can disagree in both directions on the same patient. Every laboratory threshold here is method- and laboratory-dependent, so the reader’s own laboratory’s reference interval takes precedence. A risk stratum is not a diagnosis and not a plan. This page computes the index, names the stratum and reports what it predicted in the derivation cohort; what follows is a decision for the treating team with the patient.
Frequently asked questions
What is the difference between the IPI and the R-IPI?
Only the grouping. Both score the same five factors one point each. The 1993 IPI makes four groups (0–1, 2, 3, 4–5); the R-IPI makes three (0, 1–2, 3–5). No factor, threshold or weight was changed, which is why the R-IPI could be derived without collecting new variables.
Why did the R-IPI merge the two intermediate groups?
Because in patients treated with R-CHOP their outcomes were no longer distinguishable. Rituximab compressed the middle of the distribution, so separating low-intermediate from high-intermediate stopped adding information. At the same time splitting a score of 0 out of the old low-risk group revealed a subset with 94% four-year survival.
Does the R-IPI need my laboratory’s LDH reference limit?
Only as a yes/no — the factor is ‘LDH above normal’. This page takes the measured value and the limit separately so the comparison is visible and so the ratio can be read against the NCCN-IPI, which bands it. If you only have the report’s high flag, that is enough for the R-IPI.
Should I use the R-IPI or the NCCN-IPI?
The NCCN-IPI discriminates better at both ends, but it needs the local LDH upper limit as a numeric value and uses a different extranodal item, so it is not always computable from a historical record. The R-IPI is computable wherever the 1993 index was. Compute both if you can, and say which you are quoting — the two use overlapping group names for different score ranges.
Why are the R-IPI’s survival figures so much better than the IPI’s?
Treatment era, almost entirely. The 1993 percentages come from patients on anthracycline-based chemotherapy without rituximab; the R-IPI’s come from a population-based cohort treated with R-CHOP. For the same factor count the difference is twenty to thirty percentage points. Every outcome figure here describes the cohort the index was derived in and not the patient in front of you; the spread within one stratum is wider than the gap between strata.
Related calculators
References
- Sehn LH, et al. The revised International Prognostic Index (R-IPI) is a better predictor of outcome than the standard IPI for patients with diffuse large B-cell lymphoma treated with R-CHOP. Blood. 2007. The derivation paper: a British Columbia population-based R-CHOP cohort.
- The International Non-Hodgkin’s Lymphoma Prognostic Factors Project. A predictive model for aggressive non-Hodgkin’s lymphoma (the International Prognostic Index). N Engl J Med. 1993. The parent index of both the R-IPI and the NCCN-IPI.
- University of Florida Division of Hematology & Oncology. Lymphoma prognostic indices and outcome tables: IPI, R-IPI and FLIPI.
- Counsell N, et al. Prognostic scores in diffuse large B-cell lymphoma: outcomes by IPI and R-IPI group in 1,080 patients treated with R-CHOP, with the NCCN-IPI left unassessable for want of local LDH reference limits.
- Lymphoma prognostic-index teaching deck, read for the original IPI’s 5-year overall survival by risk group (73, 51, 43 and 26%). Its R-IPI figures are read off a Kaplan–Meier curve and are NOT used.
- LaCasce AS. The International Prognostic Index: still relevant 30 years later. Haematologica. 2023;108:1453–4. Landmark Paper in Hematology series.
Not medical advice. For healthcare professionals and education. Reference intervals vary by laboratory and assay — always use your own laboratory's. Never base a dose or a treatment decision on this page alone. Full disclaimer at calcengines.com/disclaimer/