Evidence review

This evidence review is based on a published phase III study that compared the subcutaneous and intravenous formulations of trastuzumab in women with HER2-positive, non-metastatic clinical stage I–III breast cancer (HannaH study; Ismael et al. 2012; see table 1).

  • Design: 81-centre, open-label, non-inferiority randomised controlled trial. Review of pathological tumour assessment results was done by a medical reviewer masked to the women's treatment allocation.

  • Population: 596 women (18 years or over, mean 50 years) with newly diagnosed, non-metastatic, HER2-positive, primary invasive breast cancer (clinical stage I to IIIC). About half of the women had negative oestrogen receptor status and about half had cancer of clinical stage T1b–T3, N0–N1.

  • Intervention and comparison: subcutaneous trastuzumab (fixed dose of 600 mg in a volume of 5 ml plus 10,000 units rHuPH-20 [hyaluronidase]) injected into the thigh with a hand-held syringe by the nursing team at a steady rate over about 5 minutes at alternating sites every 3 weeks, compared with intravenous trastuzumab (8 mg/kg loading dose, 6 mg/kg maintenance dose) given every 3 weeks. Concurrently, all women received 8 cycles of chemotherapy: 4 cycles of docetaxel (75 mg/m2) every 3 weeks followed by 4 cycles of fluorouracil (500 mg/m2), epirubicin (75 mg/m2) and cyclophosphamide (500 mg/m2) every 3 weeks. After chemotherapy, surgery was performed according to local practice. In the adjuvant phase, radiotherapy and hormonal therapy were administered as per local practice. Trastuzumab was continued after surgery until women had completed 18 cycles of treatment.

  • Outcomes: antitumour efficacy and pharmacokinetics were co-primary outcomes, analysed in the per-protocol population. Efficacy was assessed using the pathological complete response (pCR: the absence of invasive neoplastic cells in the breast; remaining ductal carcinoma in situ was accepted). The primary pharmacokinetic outcome was pre-dose serum concentration of trastuzumab (Ctrough) recorded at cycle 8 before surgery. The primary analyses were done when all women had completed surgery (unless prematurely withdrawn) and at least 100 women in each study group had completed 1 year of treatment. The median duration of follow-up was 12.2 months (range 1.0–20.8) in the intravenous group and 12.4 months (range 0.3–20.4) in the subcutaneous group. In each group, 116 women had completed full treatment at the time of the published analysis.

Table 1 Summary of the HannaH study

Subcutaneous trastuzumab

Intravenous trastuzumab

Analysis

Randomised

n=297

n=299

Co-primary outcomes

Efficacy: per protocol set a

n=260

n=263

Primary efficacy outcome, pCRb

45.4% (118/260)

40.7% (107/263)

Treatment difference: 4.7% (95% CI −4.0 to 13.4%)

Non-inferiority criteria met: lower limit of the 2-sided 95% CI for the difference greater than pre-specified non-inferiority margin of −12.5%

Efficacy: Intention-to-treat set c

n=294

n=297

Primary efficacy outcome, pCRb

42.2%

(124/294)

37.4%

(111/297)

Statistical analysis not reported

Pharmacokinetics: per protocol set d

n=234

n = 235

Primary pharmacokinetic outcome: pre-surgery geometric mean serum Ctrough (microgram/ml; percentage coefficient of variation)

69 (55.8%)

51.8 (52.5%)

Geometric mean ratio: 1.33 (90% CI 1.24 to 1.44)

Non-inferiority criteria met: lower limit of the 2-sided 90% CI of the geometric mean ratio Ctrough subcutaneous/ Ctrough intravenous was greater than the pre-specified margin of 0.8 or more

Selected secondary outcomes, per protocol set:

Total pathological complete responsee

39.2% (102/260)

34.2% (90/263)

Treatment difference: 5.0% (95% CI −3.5 to 13.5)

Median time to response (weeks)f

6.0 (2 to 28) n=225

6.0 (3 to 25) n=231

Safety outcomes: safety analysis set g

n=297

n=298

Women reporting ≥1 adverse event (any grade)

97.3% (289/297)

93.9% (280/298)

p value for comparison not reported

Women reporting ≥1 adverse event (grade 3–5)h

51.9% (154/297)

52.0% (155/298)

p value for comparison not reported

Women reporting ≥1 serious adverse event

20.9% (62/297)

12.4% (37/298)

p value for comparison not reported

Women with adverse events leading to death

1.0%

(3/297)

<1%

(1/298)

p value for comparison not reported

Abbreviations: CI, confidence interval; pCR, pathological complete response (absence of residue invasive disease in the breast; remaining ductal carcinoma in situ was accepted).

a Subset of the intention-to-treat population with the following exclusions: women with less than 8 cycles of trastuzumab/chemotherapy treatment, metastatic breast cancer before study entry, previous anticancer therapy, HER2-negative breast cancer.

b Clinical tumour response was assessed by ultrasound or clinical examination at baseline and on day 1 of cycles 3, 5, and 7, and before surgery.

c All women with at least 1 efficacy assessment after first study drug administration.

d All women randomised and adherent to the protocol.

e Absence of invasive neoplastic cells in the breast and ipsilateral lymph nodes.

f The time from first drug administration to the date of the first clinical complete or partial response.

g All women who received at least 1 dose of trastuzumab: includes information from neoadjuvant and adjuvant phases.

h Graded according to the US National Cancer Institute's common terminology criteria for adverse events (v3.0), from 1 (mild), through 3 (severe) to 5 (fatal).

Clinical effectiveness

The HannaH study evaluated the efficacy and pharmacokinetics of 2 formulations of trastuzumab as neoadjuvant therapy in addition to chemotherapy. Subcutaneous trastuzumab was non-inferior to intravenous trastuzumab for the primary efficacy outcome of pathological complete response in the per-protocol analysis. This outcome was defined as the absence of invasive neoplastic cells in the breast; remaining ductal carcinoma in situ was accepted.

Subcutaneous trastuzumab was also non-inferior to intravenous trastuzumab for the primary pharmacokinetic outcome of pre-dose serum concentration at cycle 8 (before surgery) in the per-protocol analysis.

Among the secondary outcomes, there was no significant difference between the treatment groups for total pathological complete response (defined as the absence of invasive neoplastic cells in the breast and ipsilateral lymph nodes), and the median time to show a response to treatment (defined as the time from first drug administration to the date of the first clinical complete or partial tumour response) in the per-protocol analysis. Event-free survival and overall survival were described as secondary outcomes but data on these outcomes were not reported; the authors stated that the data were immature at the time of the analysis.

Safety

Statistical analysis of safety data was not undertaken which limits the interpretations which can be drawn.

The severity of adverse events was graded according to the US National Cancer Institute's common terminology criteria for adverse events (v3.0), from 1 (mild), through 3 (severe) to 5 (fatal). Similar numbers of women in each group experienced an adverse event of any grade of severity. The most common adverse events were alopecia, nausea, neutropenia, diarrhoea, asthenia and fatigue.

The same proportion of women in each group (52%) experienced at least 1 adverse event of grade 3 (severe) or worse. There were numerically more grade 3 adverse events and grade 4 adverse events in the intravenous group compared to the subcutaneous group (273 versus 254, and 81 versus 73, respectively) but the number of women experiencing the different grades of adverse events was not stated. Most grade 3 or worse adverse events were haematological toxic effects (most commonly neutropenia, leucopenia, and febrile neutropenia), followed by gastrointestinal disorders: the pattern of adverse events of grade 3 or worse appears similar between the study groups. Four adverse events were fatal (grade 5 severity): 1 in the intravenous group and 3 in the subcutaneous group, all of which occurred during the pre-surgery neoadjuvant phase. Of these 3 deaths, 2 were considered potentially treatment related: 1 from a myocardial infarction and 1 from septic shock.

In addition to grading by severity as above, an adverse event was classed as serious if it was fatal, life-threatening, required inpatient hospital admission or extension of existing hospital admission, resulted in persistent or significant disability or incapacity, was a congenital anomaly or birth defect, was medically significant, or required intervention to prevent any of these outcomes.

More women in the subcutaneous group than in the intravenous group had serious adverse events: new or extended hospital admissions were by far the most common reasons for an adverse event to be classified as serious. Data presented as a graph in the appendix published on line suggest that a greater proportion of serious adverse effects in the subcutaneous group than in the intravenous group related to hospital admission or prolonged stay, whereas a greater proportion of serious adverse effects in the intravenous group than the subcutaneous group were noted as being medically significant (actual numbers not reported).

The proportion of grade 3 adverse events classified as serious was numerically greater in the subcutaneous than in the intravenous group (46/254 grade 3 events [18.1%] compared with 21/273 grade 3 events [7.7%]). This was also the case for grade 4 (life-threatening or disabling) events (21/73 grade 4 events [28.8%] compared with 19/81 grade 4 events [23.5%]). In the subcutaneous group, 8.1% of serious adverse events related to 'infections and infestations' compared with 4.1% of serious adverse events in the intravenous group. However, the authors state that no infection was associated with a subcutaneous trastuzumab injection site. The most common serious adverse events were febrile neutropenia and neutropenia.

The summary of product characteristics states that cardiotoxicity (heart failure) at New York Heart Association (NYHA) class II–IV is a common adverse reaction associated with the use of intravenous trastuzumab, occurring in up to 2% of people. It states that all candidates for treatment with trastuzumab should undergo a baseline cardiac assessment. In the HannaH study, cardiac function was monitored by echocardiography or multiple-gated acquisition scan and electrocardiogram every 3 months. No cases of symptomatic congestive heart failure (NYHA class III or IV) were reported. In the subcutaneous trastuzumab group, 2 women developed congestive heart failure NYHA class II, and both had pre-existing risk factors (obesity and hypertension). The authors state that the rate of symptomatic congestive heart failure was in line with previous reports for patients with early operable or locally advanced HER2-positive breast cancer treated with trastuzumab concurrently with an anthracycline.

The authors report that the safety profiles of the intravenous and subcutaneous formulations were similar and consistent with the known safety profile of intravenous trastuzumab. See the section of this evidence summary on the evidence strengths and limitations for a discussion of the safety data. Safety and tolerability of subcutaneous trastuzumab as adjuvant therapy after surgery in women with early HER2-positive breast cancer is being assessed in the open label, non-randomised SafeHer trial (ClinicalTrials.gov identifier: NCT01566721).

Evidence strengths and limitations

The study design and analysis of results demonstrated non-inferiority of subcutaneous trastuzumab compared with intravenous trastuzumab in early (non-metastatic) breast cancer, in terms of a surrogate antitumour efficacy measure and a pharmacokinetic outcome measure. It should be noted that the treatment schedule used in the study included 24 weeks of neo-adjuvant (that is, before surgery) treatment with trastuzumab, and the pharmacokinetic outcome measure was the trastuzumab Ctrough level recorded at the end of this neoadjuvant period. Specialists have said that this is different from normal UK practice where surgery is done first unless downstaging is required or preferred.

The analysis was based on the per-protocol data set. The surrogate efficacy outcome appeared similar in the intention-to-treat analysis, but statistical analysis was not reported. The European Medicines Agency states that non-inferiority studies should be analysed using both the per-protocol and the intention-to-treat data sets.

The authors of the HannaH study note that although pathological complete response has been associated with long-term clinical benefit, there has yet to be a formal validation of it as a surrogate outcome. The lack of available data on event-free and overall survival measures means that the efficacy of the subcutaneous formulation of trastuzumab has not been proven for patient-oriented outcomes. In addition, the HannaH study relates only to women with early breast cancer, whereas the proposed marketing authorisation also includes women with metastatic breast cancer. No phase III trials which compare the intravenous and subcutaneous formulations of trastuzumab in women with metastatic breast cancer have been conducted (Roche Products Limited: personal communication January 2013).

The same proportion of women in both study groups experienced at least 1 severe (grade 3) or worse adverse events. There were numerically more grade 3 and grade 4 events in the intravenous group than in the subcutaneous group, but those that occurred in the subcutaneous group were more likely to be classed as serious. HannaH had an open-label study design because the different delivery methods of intravenous and subcutaneous administration prevented blinding. The study authors argue that awareness of treatment assignment may have affected the investigators' management of adverse events, leading them to take a more cautious approach with the investigational product and be more likely to admit women receiving the subcutaneous formulation of trastuzumab to hospital compared with those receiving the intravenous formulation. They imply that this explains why similar proportions of women in both study groups had 'severe' adverse events but there were more 'serious' adverse events in the subcutaneous group. However, alternative explanations include the possibility that admission to hospital was clinically justified more often in women in the subcutaneous group.